This article was downloaded by:[Swets Content Distribution]On: 8 May 2008Access Details: [subscription number 768307933]Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Assessment in Education: Principles,Policy & PracticePublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713404048

Assessment and Classroom LearningPaul Black a; Dylan Wiliam a

a School of Education, King's College London, London, SE1 8WA, UK

Online Publication Date: 01 March 1998

To cite this Article: Black, Paul and Wiliam, Dylan (1998) 'Assessment andClassroom Learning', Assessment in Education: Principles, Policy & Practice, 5:1,7 — 74

To link to this article: DOI: 10.1080/0969595980050102URL: http://dx.doi.org/10.1080/0969595980050102

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expresslyforbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will becomplete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with orarising out of the use of this material.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment in Vol. 5, 1, 1998

Assessment and m gL & N

School of College London, Cornwall House, Waterloo London 8WA,

T This article is a review of the literature on classroom formative assessment.Several studies show firm evidence that innovations designed to strengthen the frequentfeedback that students receive about their learning yield substantial learning gains. Theperceptions of students and their role in self-assessment are considered alongside analysisof the strategies used by teachers and the formative strategies incorporated in suchsystemic approaches as mastery learning. There follows a more detailed and theoreticalanalysis of the nature of feedback, which provides a basis for a discussion of thedevelopment of theoretical models for formative assessment and of the prospects for theimprovement of practice.

nOne of the outstanding s of studies of assessment in t s has been theshift in the focus of attention, s t in the s betweenassessment and m g and away m n on the sof d s of test which e only weakly linked to the g sof students. This shift has been coupled with many s of hope that

t in m assessment will make a g n to thet of . So one main e of this w is to y the

evidence which might show whethe o not such hope is justified. A second eis to see whethe the l and l issues associated with assessment fo

g can be illuminated by a synthesis of the insights g amongst the estudies that have been .

The e of this n is to y some of the key y that weuse, to discuss some s which define the baseline m which ou studyset out, to discuss some aspects of the methods used in ou , and finally to

e the e and e fo the subsequent sections.Ou y focus is the evidence about e assessment by s in thei

school o college . As will be explained below, the y fo theh s and s that have been included has been loosely than

tightly . The l n fo this is that the m e assessmentdoes not have a tightly defined and widely accepted meaning. n this , it is tobe d as encompassing all those activities n by , and/o bythei students, which e n to be used as feedback to modify the

0969-594X/98/010007-68 ©1998 x g Ltd

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 8 . Black & D. Wiliam

teaching and g activities in which they e engaged.Two substantial w , one by o (1987) and the othe by s

(1988) in this same field e as baselines fo this . , with a fewexceptions, all of the s d e e published g o afte 1988. The

e h was conducted by l means. One was h a citation hon the s by o and , followed by a simila h on late and

t s of component issues published by one of us , 1993b), and bys and the k et al., 1990; s et al.,

1991a,b). A second h was to h by s in the C data-base;this was an inefficient h because of a lack of s used in a m waywhich define ou field of . The d h was the 'snowball' hof following up the e lists of s found. Finally, fo 76 of the most likely

, the contents of all issues e scanned, m 1988 to the t in somecases, m 1992 fo s because the k had y been done fo the 1993

w by k (see Appendix fo a list of the s scanned).s w d a field than ou own. The pape spanned a full

e of assessment , which he d as , selection,n and motivation. Only the last two of these e d . s used

the m m evaluation' with the same meaning as we e fo eassessment'. These two s gave e lists containing 91 and 241 items

, but only 9 items appea in both lists. This s the twin andd difficulties of defining the field and of g the .

The s of composing a k fo a w e also d by thes between the o and the s . o s the

issues within a k d by a model of the assessment cycle, which sm , then moves to the setting of tasks, a and , then

h to g e and g feedback and outcomes. e thendiscusses h on the impact of these evaluation s on students. shis most significant point, , is that in his view, the vast y of the

h into the effects of evaluation s is t because key distinctionse conflated (fo example by not g fo the quality as well as the quantity

of feedback). e concludes by suggesting how the weaknesses in the existinge might be d in e .

' pape has a focus—the impact of evaluation s onstudents—and divides the field into e main e impact of l class-

m testing , the impact of a e of othe l s whichbea on evaluation, and finally the motivational aspects which e to mevaluation. e concludes that the summative function of sbeen too dominant and that e emphasis should be given to the potential of

m assessments to assist . Feedback to students should focus on thetask, should be given y and while still , and should be specific to thetask. , in ' view the 'most vital of all the messages g m this

' (p. 470) is that the assessments must emphasise the skills, knowledge andattitudes d to be most , howeve difficult the technical sthat this may cause.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 9

Like s , the h cited by s s a e of styles andcontexts, d studies involving k in l s bythe students' own , to s in y settings by . The

e of k that is not d out in l s by s can becalled in question g & Fox, 1991), but if all such k e excluded, notonly would the field be y populated, but one would also be gmany t clues and s s the difficult goal of g an ade-quately complex and complete g of e assessment. Thus this

, like that of o and e y that of , is eclectic. nconsequence, decisions about what to include have been somewhat , so thatwe now have some sympathetic g of the lack of p between the

e s used in the two .The s d above d a total of 681 publications which

d , at sight, to the . The c details fo thoseidentified by c means e d (in most cases, including )into a c database, and the s e d manually. An initial

, in some cases based on the t alone, and in some cases involvingg the full publication, identified an initial total of about 250 of these publica-

tions as being sufficiently t to e g in full. Each of thesepublications was then coded with labels g to its focus—a total of 47 tlabels being used, with an e of 2.4 labels pe . Fo each of thelabelled publications, existing s e d and, in some cases modifiedto highlight aspects of the publication t to the t , and s

n e none existed in the database. d on a y g of thet , a e of seven main sections was adopted.

The g fo each section was n by allocating each label to asection. All but one of the labels was allocated to a unique section (one was allocatedto two sections). s of publications t to each section e then dout togethe and each section was allocated to one of the s so that initial scould be , which e then d jointly. The seven sections which

d this s may be y d as follows.The h in the section on Examples in evidence is , in that an

account is given of a y of selected pieces of h about the effectivenessof e assessment, and then these e discussed in to identify a set of

s to be e in mind in the e analytic—sections.The next section on Assessment by s adds to the l dby g a f account of evidence about the t state of eassessment e amongst .

e follows a e d account of the field. The next two sections dealy with the student e and the ' . Whilst the section

on s and tactics fo s focuses on tactics and s in , thenext section on Systems follows by discussing some specific and esystems fo teaching in which e assessment plays an t . Thesection on Feedback is e e and , g an account,

d in evidence, of the e of feedback, a concept which is l to

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 10 Black & D. Wiliam

e assessment. This s the d fo a final section, on s fothe y and e of e assessment, in which we attempt a synthesis ofsome of the main issues in the context of an attempt to w the l basis,the h s and needs, and the implications fo e and fo policy of

e assessment studies.

Examples in EvidenceClassroom

n this section we t f accounts of pieces of h which, between ands them, e some of the main issues involved in h which aims toe evidence about the effects of e assessment.

The first is a t in which 25 e s of mathematics e din self-assessment methods on a 20-week e , methods which they putinto e as the e d with 246 students of ages 8 and 9 and with108 olde students with ages between 10 and 14 (Fontana & , 1994). Thestudents of a 20 e s who e taking anothe e ineducation at the time d as a l . h l and l

s e given - and post- tests of mathematics achievement, and both spentthe same times in class on mathematics. h s showed significant gains ovethe , but the l s mean gain was about twice that of the

l s fo the 8 and d students—a y significant .Simila effects e obtained fo the olde students, but with a less clea outcomestatistically because the , being too easy, could not identify any possibleinitial e between the two . The focus of the assessment k was on

y daily—self-assessment by the pupils. This involved teaching themto d both the g objectives and the assessment , giving them

y to choose g tasks and using tasks which gave them scope to assessthei own g outcomes.

This h has ecological validity, and gives d evidence ofg gains. The s point out that e k is d to look fom outcomes and to e the e effectiveness amongst the l

techniques employed in . , the k also s that an initiativecan involve fa e than simply adding some assessment s to existingteaching—in this case the two outstanding elements e the focus on self-assessmentand the implementation of this assessment in the context of a t class-

. On the one hand it could be said that one o othe of these , o thecombination of the two, is e fo the gains, on the othe it could be dthat it is not possible to e e assessment without some l changein m pedagogy because, of its , it is an essential component of thepedagogic .

The second example is d by Whiting et al. (1995), the autho being theteache and the s y and school t staff. The account is a

w of the s e and , with about 7000 students ove a

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 11

d equivalent to 18 , of using y g with his classes. Thisinvolved testing and feedback to students, with a t that theyeithe achieve a high test t least e they e allowed to dto the next task, , if the e e , they study the topic until theycould satisfy the y . Whiting's final test s and the e point

s of his students e consistently high, and highe than those of students inthe same e not taught by him. The students' g styles e changed asa t of the method of teaching, so that the time taken fo successive units was

d and the s having to e tests . n addition, tests ofthei attitudes s school and s g showed positive changes.

Like the s study, this k has ecological validity—it is a t of k inl s about what has become the l method used by a teache ove

many . The gains d e substantial; although the s with thel e not documented in detail, it is d that the teache has had difficulty

explaining his high success e to colleagues. t is conceded that the success couldbe due to the l excellence of the , although he believes that the

h has made him a bette . n he has come to believe thatall pupils can succeed, a belief which he s as an t t of the

. The t shows two c and d e beingthat the teaching change involves a completely new g e fo the students,not just the addition of a few tests, the second being that y because of this,it is not easy to say to what extent the effectiveness depends specifically upon thequality and communication of the assessment feedback. t s theexample in g movement aimed at a l change in g

, and in that it is based on t assumptions about the e of.

The third example also had its n in the idea of y , but d the y in that the s d the belief that it was the

testing that was the main cause of the g achievements d fothis . The t was an t in mathematics teaching z &

, 1992), in which 120 n college students in an y ae e placed in one of fou s in a 2 X 2 l design fo an

18-week e g seven s of a text. Two s e given one testpe , the othe two e given e tests pe . Two s etaught by a y d and highly d , the othe two by a y

d teache with e . The s of a post-test showed asignificant advantage fo those tested e but the gain was fa smallefo the d teache than fo the . n of the final swith the p of students in the same e but not in the tshowed that the d teache was indeed exceptional, so that the scould conclude that the e testing was indeed effective, but that muchof the gain could be d by an exceptional teache with less testing.

y n with the study above, this one has simila statistical sand analyses, but the e of the two s being d is quite .

, one could question whethe the testing y constitutes e

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 12 Black & D. Wiliam

assessment—a discussion of that question would have to focus on the quality of thet n and on whethe test s constituted feedback in the

sense of leading to e action taken to close any gaps in e, 1983). t is possible that the y of the d teache

may have been in his/he skill in this aspect, thus making the testing e effectivelye at eithe .

Example numbe four was n with d n being taught inn n et al., 1991). The g motivation was a belief that close

attention to the y acquisition of basic skills is essential. t involved 838 nn mainly m disadvantaged home s in six t s in the

USA. The s of the l p e d to implement a -ment and planning system which d an initial assessment input to mteaching at the individual pupil level, consultation on s afte two weeks, newassessments to give a diagnostic w and new decisions about students'needs afte fou weeks, with the whole e lasting eight weeks. The s usedmainly s of skills to assess , and d with open-style activitieswhich enabled them to e the tasks within each activity in to matchto the needs of the individual child. e was emphasis in thei g on a

d model of the development of g n up on thebasis of s of , and the diagnostic assessments e designed to helplocate each child at a point on this scale. Outcome tests e d with initialtests of the same skills. Analysis of the data using l equation modellingshowed that the t s e a g t of all outcomes, butthe l p achieved significantly highe s in tests in ,mathematics and science than a l . The n tests used, which e

l multiple-choice, e not adapted to match the open d styleof the l s . , of the l , on e 1child in 3.7 was d as having g needs and 1 in 5 was placedin special education; the g fo the l p e 1 in17 and 1 in 71.

The s concluded that the capacity of n is d inconventional teaching so that many e 'put down' y and so have thei

s . One e of the s success was that s hadenhanced confidence in thei s to make l decisions wisely. This example

s again the embedding of a e assessment e within aninnovative . What is e salient e is the basis, in that , ofa model of the development of e linked to a n based scheme ofdiagnostic assessment.

n example numbe five , 1988), the k was d e y inan explicit psychological , in this case about a link between c motiv-ation and the type of evaluation that students have been taught to expect. The

t involved 48 d i students selected m 12 classes s4 schools, half of those selected being in the top e of thei class on tests ofmathematics and language, the othe half being in the bottom . The students

e given two types of task in , not m , one of each pai testing

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 13

t thinking, the othe . They e given n tasks to be tackledindividually unde , with an l n and explanation. esessions e held, with the same pai of tasks used in the t and . Eachstudent d one of e types of n feedback with d , both onthe session's k e the second, and on the second session's k ethe . The second and d sessions, including all of the t and non the feedback, d on the same day. Fo feedback, d of the p

e given individually composed comments on the match, o not, of thei kwith the a which had been explained to all . A second p egiven only , d the s on the g session's . The

d p e given both s and comments. s on the k done in eachof the e sessions d as outcome . Fo the 'comments only' pthe s d by about d between the and second sessions, foboth types of task, and d at this highe level fo the d session. The'comments with ' p showed a significant decline in s s the esessions, y on the t task, whilst the e only' p declinedon both tasks between the and last sessions, but showed a gain on the secondsession, in the t task, which was not subsequently maintained. Tests ofpupils' t also showed a simila : , the only significant -ence between the high and the low achieving s was that t was -mined fo the low s by eithe of the s involving feedback of ,

s high s in all e feedback s maintained a high level of.

The s e discussed by the s in s of cognitive evaluation .A significant e e is that even if feedback comments e y helpfulfo a student's , thei effect can be d by the negative motivationaleffects of the e feedback, i.e. by giving . The s e consistentwith e which indicates that task-involving evaluation is e effective thanego-involving evaluation, to the extent that even the giving of e can have anegative effect with . They also t the view that nwith e attainment can lowe the quality of task , y on

t tasks.This study s two significant messages fo this l . The is

that, whilst the t lacks ecological validity because it was not t of od to l m k and was not d out by the students' usual

, it s might e some t lessons about ways in whiche evaluation feedback might be made e o less effective in lm . The second lesson is the possibility that, in l m ,

the effectiveness of e feedback will depend upon l detailed s ofits quality, and not on its e existence o absence. A d message is that closeattention needs to be given to the l effects between low and high ,of any type of feedback.

The sixth example is in l ways simila to the fifth. n this k (Schunk,1996), 44 students in one USA y school, all 9 o 10 s of age, dove seven days on seven packages of l s on unde the

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 14 Black & D. Wiliam

s of e students. Students d in fou e s subjectto t two s the s d g goals nhow to solve ) whilst fo the othe two they d e goals

y solve them). Fo each set of goals, one p had to evaluate theig capabilities at the end of each of the sessions, s the othe

was asked instead to complete an attitude e about the . Outcomes of skill, motivation and self-efficacy showed that the p given -

ance goals without self-evaluation came out lowe than the othe e on all. The n of this t suggested that the effect of the t

self-evaluation had out-weighed the l effect of the two types of goal. Thiswas d in a second study in which all students k the self-evaluation,but on only one occasion nea the end than afte all of the six sessions.

e e two s who d only in the types of goal that e empha-sised—the aim being to allow the goal effects to show without the possible -whelming effect of the t self-evaluation. As expected, the g goal

n led to highe motivation and achievement outcomes than did thee goal.

The k in this study was m , and the s given in allfou ' e of types that might have been given by t ,although the high y of the self-evaluation sessions would be y unusual.Thus, this study comes close to ecological validity but is s an t

d outside l class conditions. t s with the s (fifth) studythe focus on goal , but shows that this e s with evaluativefeedback, both within the two types of task, and whethe o not the feedback is

d m an l e o m self-evaluation.The seventh example involved k to develop an d middle school

science-based m n & White, 1997). The teaching e wasfocused on a l y h to g about e and motion, and the

k involved 12 classes of 30 students each in two schools. Each class was taughtto a y d m plan in which a sequence of conceptuallybased issues was d h s and compute simulation, using an

y cycle model that was made explicit to the students. All of the k wasd out in pee . Each class was divided into two halves: a l p

used some s of time fo a l discussion of the module, whilst anl p spent the same time on discussion, d to e

e assessment, with both pee assessment of s to the class andself-assessment. This l k was d d students' use oftools of systematic and d , and the social context of g and othecommunication modes. All students e given the same basic skills test at theoutset. The outcome s e of e types: one a mean e on s

t the , one a e on two chosen s which each studentd out independently, and one a e on a conceptual physics test. On the

mean t , the l p showed a significant l gain;, when the students e divided into e s g to low,

medium o high s on the initial basic skills test, the low g p showed

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 15

a , ove thei l p , of e than e d deviations,the medium p just ove two, and the high p just ove one. A simila ,of y of the l p which was e d fo low gstudents on the basic skills test, was also found fo the othe two outcomes. Amongstthe students in the l , those who showed the best gof the assessment s achieved the highest .

This science t again shows a n of e assessment which is anc component of a e g innovation to change teaching and. Whilst the l e e lay only in the develop-

ment of e assessment' amongst the students, this k was embedded in ant e such assessment was an c component. Two othe distinc-

tive s of this study e the use of outcome s of t types,but all y g the aims of the teaching, and second the l gainsbetween students who would have been labelled 'low ability' and 'high ability'

.The eighth and final example is t m the , in that it was a

meta-analysis of 21 t studies, of n g m l to e12, which between them yielded 96 t effect sizes (Fuchs & Fuchs, 1986).The main focus was on k fo n with mild handicaps, and on the use of thefeedback to and by . The studies e y selected—all involved

n between l and l , and all involved assessmentactivities with s of between 2 and 5 times pe week. The mean effect sizeobtained was 0.70. Some of the studies also included n without handicap:these gave a mean effect size of 0.63 ove 22 sets of s (not significantly t

m the mean of 0.73 fo the handicapped . The s noted that in abouthalf of the studies s d to set s about s of the data and actionsto follow, s in the s actions e left to ' judgments. The

d a mean effect size of 0.92 d with 0.42 fo the . ,those studies in which s k to e s of the s ofindividual n as a guide and stimulus to action d mean gains thanthose e this was not done (mean effect size 0.70 d with 0.26).

e s of this last example e of t . The is thatthe s e the g success of the e h with the

y outcomes of s which had attempted to k is fo individualised g s fo , based on -

la g s and diagnostic . Such s embodied a deduc-tive h in t with the inductive h of e feedback

. The second e is that the main g gains the ek e only achieved when s e d to use the data in

systematic ways which e new to them. The d e is that such accumula-tion of evidence should have given some l impetus to the development of

e assessment—yet this pape s to have been d in most of thelate .

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 16 Black & D. Wiliam

Some General

The studies chosen thus fa e all based on quantitative s of ggains, six of them, and those d in the eighth, being in using - andpost-tests and n of l with l . We do not implythat useful n and insights about the topic cannot be obtained by k inothe .

As mentioned in the , the ecological validity of studies is yt in g the applicability of the s to l m ., we shall assume that, given this caution, useful lessons can be t m

studies which lie at s points between the ' m and the specialconditions set up by . n this t all of the studies exhibit some eof movement away m ' . The study (by Whiting et al., 1995)which is most y one of l teaching within the y m is,inevitably, the one fo which quantitative n with a y equivalent

l was not possible. e , caution must be d fo any studiese those teaching any l s e not the same s as those fo

any l .Given these , , it is possible to e some l

s which these examples e and which will e as a k fo latesections of this .

t is d to see how any innovation in e assessment can be d as al change in m . All such k involves some e of

feedback between those taught and the , and this is entailed in the qualityof thei s which is at the t of pedagogy. The e of these

s between s and students, and of students with one , willbe key s fo the outcomes of any changes, but it is difficult to obtaindata about this quality m many of the published . The examples doexhibit t of the y of ways in which enhanced e k can beembedded in new modes of pedagogy. n , it can be a salient and explicit

e of an innovation, o an adjunct to some t and scalemovement—such as y . n both cases it might be difficult to

e out the n of the e feedback to any ggains. Anothe evaluation m that s e is that almost all innovations

e bound to be g innovations in ends as well as in means, so that thedemand fo unambiguous quantitative s of effectiveness can neve befully satisfied.

g the s s e assumptions about the psychology of. These can be explicit and fundamental, as in the t basis of

the t and the last of the examples, o in the diagnostic h of n etal. (1991) o implicit and , as in the y g .

Fo assessment to be e the feedback n has to be used—whichmeans that a significant aspect of any h will be the l swhich e d in e to the feedback. e again assumptions

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 17

about , and about the e and e of g tasks which wille the best challenges fo d , will be significant. The -

ent s and s s these assumptions e the possibility of a widee of s involving e assessment.

The e of students in assessment is an t aspect, hidden because it istaken fo d in some , but explicit in , y e selfand pee assessments by and between students e an t e (withsome g that it is an inescapable e , 1989).

The effectiveness of e k depends not only on the content of thefeedback and associated g , but also on the context ofassumptions about the motivations and s of students within whichit . n , feedback which is d to the objective needs ,with the assumption that each student can and will succeed, has a y teffect that feedback which is subjective in mentioning n with

, with the assumption—albeit t some students e not as able ass and so cannot expect full success.

, the consistent e s the of these examples is that they allshow that attention to e assessment can lead to significant g gains.Although e is no e that it will do so e of the context and the

h adopted, we have not come s any t of negative effectsfollowing on an enhancement of e . n this , one lmessage of the s w has been .

One example, the n study of n et al. (1991) s out -cally the e that may be attached to the achievement of such gains. This

innovation has changed the life chances of many . This py may not look as t as it y is when a t is d y ins of effect sizes of (say) 0.4 d deviations.

To glean e the published , it is y to change gea and moveaway holistic s of selected examples to a e analytic m of

. This will be n in the next five sections.

Assessment by TeachersCurrent

' s in e assessment e d in the s by s(1988) and k (1993b). l common s d these .The l e was one of weak . y weaknesses :

m evaluation s y e l and e ,g on l of isolated details, usually items of knowledge which pupils

soon .s do not y w the assessment questions that they use and do

not discuss them y with , so e is little n on what is beingassessed.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 18 . Black & D. Wiliam

The g function is d and the g function -emphasised.

e is a tendency to use a e than a n , whichemphasises competition between pupils than l t ofeach. The evidence is that with such s the effect of feedback is to teach theweake pupils that they lack ability, so that they e de-motivated and loseconfidence in thei own capacity to .

e t h has d this l . s appea to bee of the assessment k of colleagues and do not t o use thei

assessment s (Cizek et al, 1995; l et al, 1997). h in questioning andn , ' assessment focuses on low-level aims, mainly . e is

little focus on such outcomes as speculation and l n (Stiggins et al,1989; Schilling et al., 1990; , 1992; l & , 1996; Senk et al, 1997), andstudents focus on getting h the tasks and t attempts to engage incognitive activities l & , 1997). Although s can t the

e of thei pupils on l tests—albeit tests g low-level aims—thei own assessments do not tell them what they need to know about thei students'

g h et al, 1992; , 1987).s of y school s in England and in e have d that

' s tend to emphasise the quantity of students' k than itsquality, and that whilst tasks e often d in cognitive , the assessments ein affective , with emphasis on social and l functions t et al,1992; d et al., 1994; , 1996). e e some g commentsby those who have d these issues—one t on science s sees

e and diagnostic assessment as 'being y in need of development'l et al, 1995, p. 489), anothe closes with a puzzled question 'Why is the

extent and e of e assessment in science so ' s &Singh, 1996, p. 99), whilst a y of s in Quebec , Canada, sthat fo e assessment d they pay lip e to it but conside that its

e is c in the t educational context' (quoted by a et al.,1993, p. 116). The conclusion of a y about e in n y schoolswas that the a used by s e y invalid by l '

, 1991, p. 104). A study which used s and so d ae of the s of US s concludes as follows:

t of the s in this study e caught in conflicts among beliefsystems, and institutional , agendas, and values. The point of

n among these conflicts was assessment, which was associated withy l feelings of being , and of , guilt,

, and . These s d difficulty in keeping kof and having the language to talk about s e development.They also d e l accountability testing. They

d in thei assessment s and in the language they used toe students' y development. Those who d in highly

g situations e inclined to use blaming language and tended to

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 19

e global, negative e assessments in l language.Thei assessments e likely to be based on a simple, linea notion of

. The less g the situation the less this was likely to .This study suggests that assessment, as it s in schools, is fa m a

technical . , it is deeply social and . John-ston et al., 1995, p. 359)

This last quotation also s attention to the dominance of l summativetesting. The effects e n deep, witness the evidence in n that when s

e d to e thei own assessments they imitated the l testst et al., 1992), and seemed to be able to think only in s of t

summative tests with no feedback action , 1992; n & ,1996). A simila effect was d in assessment s in Queensland

& , 1987). A t tension between e and summativeassessment s when s e e fo both functions: e has beendebate between those who w attention to the difficulties of combining the two

s (Simpson, 1990; Scott, 1991; n et al., 1992) and those who e that itcan be done and indeed must be done to escape the dominance of lsummative testing , 1993a; Wiliam & , 1996). The t inScotland, that s use l tests when they think thei pupils e , andmainly fo n s (i.e. checking fo consistency of s betweenschools), does not seem to have d these tensions n et al., 1995).

Assessment, and

Given these , it is not g that when national o local assessmentpolicies e changed, s become confused. l of the s quoted abovegive evidence of this. A patchy implementation is d fo s of teacheassessment in e t et al., 1996) and in h Canada , 1990),whilst in the U such changes have d a y of , some of whichmay be e and in conflict with the stated aims of the changes which

d them m et al., 1993; Gipps et al., 1997). e changes havebeen d with substantial g o as an c t of a t in which

s have been closely involved, the pace of change is slow because it is ydifficult fo s to change s which e closely embedded within theiwhole n of pedagogy , 1989; d et al., 1994, 1996; ,1995) and many lack the e s that they need to e themany e bits of assessment n in the light of d g s

& , 1994). , some such k fails to e its effect. At with s in the e , which d to n them to communicate

with students in to e the students' view of thei own , found thatdespite the , many s stuck to thei own agenda and failed to dto cues o clues m the students which could have d that agenda

, 1994).The issue that s , as it did in the section above on m

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 20 Black & D. Wiliam

, is the close link of e assessment e both with othecomponents of a s own pedagogy, and with a s conception of his ohe . n a t aimed at enhancing the powe of science s to ethei students at , s could not find time fo g because they enot d to change m s in to give students e e

y and give themselves a less closely demanding . The sd this as a e to k the existing symbiosis of mutual dependency

between s and students (Cavendish et al., 1990). n h with specialeducation , Allinde (1995) found that s with a g belief in theihigh l and teaching efficacy made bette use of e assessment thanthei less confident .

We have not d e to give a e w of the e on' assessment . The aim has been to highlight some key points which

e t to the main e of this . The e outstanding s :

that e assessment is not well d by s and is weak in;

that the context of national o local s fo n and accountabil-ity will t a l influence on its ; and

that its implementation calls fo deep changes both in ' sof thei own e in n to thei students and in thei m .

These s have implications fo h into this . h which simplys existing e can y do little e than m theg findings d above. To be e , h has to be

linked with a e of . f such n is to seek implemen-tation with and h s in thei l , it will be changing thei

s and ways of teaching; then the e initiative will be t of an of changes and its evaluation must be seen in that context. e

closely focused pieces of h might be e e as ways of g thet issues that e involved, but might have to use d s

because s cannot be expected quickly to abandon habitual s and methodsfo a limited . Thus at least some of the h that is needed willinevitably lack ecological validity.

Students and Formative AssessmentThe e of the activity of e assessment lies in the sequence of two actions.The is the n by the of a gap between a d goal and his ohe t state (of knowledge, and/o , and/o skill). The second isthe action taken by the to close that gap in to attain the d goal

, 1983; , 1989). Fo the action, the e y fog the n may lie with the student in self-assessment, o with

anothe , notably the , who s and s the gap andcommunicates a message about it to the student. Whateve the s by whichthe assessment message is , in n to action taken by the it

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 21

would be a mistake to d the student as the passive t of a call to action.e e complex links between the way in which the message is , the way

in which that n motivates a selection amongst t s of action,and the g activity which may o may not follow.

Fo the s of this , the involvement of students in e assess-ment will be d by division into two d topics, as follows:

(1) The of these will focus on those s which influence the n of themessage and the l decisions about how to d to it. The nwill be with the effects of beliefs about the goals of , about one'scapacity to , about the involved in g in s ways, andabout what g k should be like: all of these affect the motivation totake action, the selection of a line of action and the e of one's commitmentto it.

(2) The second will focus on the t ways in which positive action may betaken and the s and g contexts in which that action may be dout. The focus e will be on study methods, study skills, n with

, and on the possibilities of pee and self-assessment.

e is y a g n between the two . n , if self andt e d in a , this affects the initial n of the

message about a gap as well as the way in which a may k to close it., the g sets of beliefs to be d within the focus bea

on the n of and e to feedback messages, albeit in t ways,whethe they e d by the self o by . n the studies d withinthe topic, both s of feedback have been .

and

n his analysis of e assessment by s in , d commentsthat:

A numbe of pupils do not e to n as much as possible, but econtent to 'get by', to get h the , the day o the yea withoutany majo , having made time fo activities othe than school k[...] e assessment y s a shift in this equilib-

point s e school , a e s attitude to g[...] y teache who wants to e e assessment must recon-struct the teaching contracts so as to counteract the habits acquired by his pupils.

, some of the n and adolescents with whom he is dealinge d in the identity of a bad pupil and an opponent. ,

1991, p. 92 s italics))

This pessimistic view is , but modified, by the finding of Swain(1991) that some y students g on teache assessed science s inEngland would d to s difficulties by g on y aspects ofthe task, so avoiding the main , and would be 'insatiable' in thei h fo

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 22 Black & D. Wiliam

cues fo the t ' m . These symptoms of y eaccompanied by t moves to e the esteem of the . ,

d s (1992) that some US students will y to avoid the involvedin tackling a challenging assignment.

Thus whilst e to be n into a e s engagement with gk may e wish y to minimise , e can be othe influences.

One m may be e a l commitment d can y withit an enhanced penalty fo e in s of one's self-esteem. Anothe mmay be that students can fail to e e feedback as a helpful signal andguide (Tunstall & Gipps, 1996a). e & s (1996) e study of the

s of Japanese and n students, which aimed to e thei self-n , shows that e can be y . ys t that positive g gains d by e feedback e

associated with e positive attitudes to y in y gs e the use to be made of the feedback is y planned k et al.,

1990; Whiting et al., 1995), but e can also be negative affects and the notionsof attitude and motivation have to be d in e detail if the n of sucheffects is to be .

n the w and analysis d by d (1992), he points to evidencethat students can be t to seek help, and e not always happy to e aassistance because it is d as evidence of thei low ability. , in thei

l study of the effects of t s of guidance with d and 6ths solving mathematical , Newman & Schwage (1995) found that,

whilst the t s could make a , the of sfo help all students was y low and they concluded that e is aneed to e e help-seeking in the y . The l eof this study was that the e between the two s of feedbackguidance being given was a seemingly w one. One p e told that thegoals of the k e in learning ('This will help you to n new things...') withemphasis on the e of g how to tackle s of the type

, whilst fo the othe the goal d was thei own performance w youdo helps us to know how t you e and what kind of e you will get...') with

g emphasis on completing as many s as possible. tthis , all d the same tuition, including feedback, in t of the

k and all e d to seek fo help wheneve they felt the need. Theperformance goal students e e likely to show maladaptive questioning sand solved fewe , y when those initially classified as low

s e d s the two .

Goal Orientation

This effect of goal n on g has been extensively studied. The studyof Ames & (1988) involved only y into the goals that students yheld. They found that thei sample of 176 students g ove s 8 to 11could be divided into two e with y n and those with

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 23

e . The spoke of the e of , believedin the value of t to achieve , and had a y positive attitude to

. The latte d e to lack of ability, spoke e in s of theie ability, about g with y little t if able, and focused on the

significance of g . A simila distinction was made in the -vention study by (1988) y d in the section on m

e above in which the s 'ego-involving feedback' and 'task-involvingfeedback' e used. The g t of this study, that the giving of scould e the positive help given by task comments, s the sensitivityof the issues d . n a late study, & Neuman (1995) showed thatthose in task mode e e likely to seek help and to explain help-avoidance in

s of seeking independent , whilst those in an ego mode sought help lessand explained thei avoidance in s of masking thei incapacity. Two l

s of this field both s that feedback which s attention away thetask and s self-esteem can have a negative effect on attitudes and e

n & , 1994; & , 1996). t is even the case that givinge can have bad effects, y when it is not linked to objective feedback

about the . Leppe & l (1989) e that d systems can eboth t and motivation, whilst a detailed study by & e (1996)shows how a teache can e on e e fo a child at the expenseof helping the child to .

l studies by Schunk (Schunk, 1996) have developed this same theme. Thishas y been t out in the one d in the section on m

. n two studies, one on the g of g with 5th e lstudents (Schunk & , 1991), the othe on g n with m

s (Schunk & , 1993a), the second showed that bette s ed by giving s goals than t goals, and both showed that

e the feedback on s goals was supplemented to include n aboutstudents' s s the l aim of the , both the students'

g e and thei beliefs about thei own e capacities(self-efficacy), e at the highest level. The s of association between achieve-ment, self concept, and the s of study and feedback d by studentshave been the subject of a detailed analysis, using s 12 high schoolbiology , by Thomas et al. (1993). A complex n of links , butthe e of self-concept was , and it also seemed that the n ofchallenging assignments and extensive feedback lead to student engagementand highe achievement.

Self-perception

n a e l w of the e in this field, Ames (1992) d m theevidence about the advantages that ' (i.e. ) goals can e and

s the salient s of the g s that can help to ethese advantages. She concludes that evaluation to students should focus on individ-ual t and , but e this the tasks d should help

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 24 Black & D. Wiliam

students to establish thei own d goals by g a meaningful,g and y demanding challenge. She also s that feed-

back should be , must be linked to s fo , andshould e the view that mistakes e a t of . The nof students is t , and this will be y influenced by 'beliefs about the e e of ' as against 'ability' in thei views of

. n , it is t that motivation is seen to involve changes instudents' qualitative beliefs about themselves, which the setting of goals and the styleof feedback should both be designed to . The use of c s can be

e if they focus attention on 'ability' than on the belief thatone's t can e success. Of , the beliefs of s and of s canalso affect the ways in which the self-concepts of students e developed, as ispointed out in s analysis (1992), which s l conclusionssimila to those of Ames.

e is evidence m many studies that ' beliefs about thei owncapacity as s can affect thei achievement. Examples that can be added tothose y quoted above e those of Lan et al. (1994), n et al. (1991),

s & Fontana (1996), g (1994) and & Winne (1995). The studyof s & Fontana showed that achievements within the t in

l d in the section on m e e linked to anenhancement of the students' sense of thei own l ove thei , and

s k also focused on locus of l as a of . k& n (1987) d that d g styles d betteconceptual , an effect that they d to enhanced autonomy and

l locus of . These issues e analysed in a l pape by i& n (1994) which is discussed in the section on k .

Studies by Skaalvik (1990), o & van Oudenhoven (1995) and Vispoel &Austin (1995) all show that the s students gave fo the s of thei gdiffe between low , who e e to low ability, and high swho tend to e success to . Vispoel & Austin e that s shouldhelp students to e s to ability, and should e them to

d ability as a collection of skills that they can maste ove time.s k in mathematics and g with students in s 3 to 6 n

et al., 1991), showed that students' self-concept could be enhanced by feedbackdesigned to this end and that whilst those whose self-concept was initially lowshowed e gains, those with initially high self-concept showed no gains. naddition, the students' n of success in the k to t d whilst

s to ability did not. , in this t , the sobtained by the could not be d by the teache and e e nosignificant s in achievement between t and l . Afinal and e is added by the w of & Winne (1995), who,in addition to g the evidence that many of the s mentioned above canhave on g achievement, also w attention to the e of 'beliefs about the e of , about the amount of t that successful

g can demand, about the e of , and about —

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 25

expectation that all g should lead to simple and unambiguous s to allthe questions that can be .

, this section of this w has been selective and does not claim to covethe many possible aspects implied in the s attitude and motivation. The

focus in the k d e is to call attention to the eof a y of l , , self-efficacy, andassumptions about the e of . e e y complex s and

s between these ; n & Schmeck (1995) in ae analysis of evidence on these , have d an

y of g ' in to e what they call 'a multi-facetede on individual s in .

The e of these s s m the conjunction of two types ofh s d above. One is that the l ' d to

above can have t effects on a student's . The othe is that the wayin which e n is conveyed to a student, and the context of

m e and beliefs about ability and t within which feedback isd by the individual , can affect these l s fo good o

ill. The hopeful message is that innovations which have paid l attention tothese s have d significant g gains when d with theexisting s of m .

Assessment by Students

The focus of this section is to discuss one aspect of the g activity which mayfollow m the student acceptance and g of the need to close a gapbetween t achievement and e goals. n e assessment, anyteache has a choice between two options. The is to aim to develop the capacityof the student to e and e any gaps and leave to the student the

y fo planning and g out any l action that may be needed.This option implies the development within students of the capacity to assessthemselves, and s to e in assessing one . The secondoption is fo s to take y themselves fo g the stimulus

n and g the activity which follows. The of these two will bethe subject of this section, whilst the second will be discussed in the sectionstitled s and tactics fo s and Systems below. The two options pin that it is possible to combine the two : the y between thissection and the section on s and tactics fo s will e be

, as is the y between this section and the section on m.

The focus on self-assessment by students is not common , even amongstthose s who take assessment . s & Singh (1996) found that onlyabout a d of the U science s whom they sampled involved pupils yin thei own assessment in any way, and both n & s (in etah, 1994, pp. 15-28) and the account of n initiative by t din k & Atkin, 1996, pp. 92-119) e the n of self-assessment,

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 26 Black & D. Wiliam

y in y school science in the U and in y mathematics in, as innovations. n the l e on m assessment, the topic

is y example, the e e collection bye (1997) contains no piece which focuses explicitly on self- and .

The motives fo g this e e . n & s dbecause of the l impossibility of g the level of need of eachindividual in a class of about 30 students engaged in l y fthey could do it fo themselves the teache could deploy his/he t eefficiently. n his w of the e on student self-evaluation in l

g s in the health sciences, (1995) d that the eskills e not y taught in most , but also d new

h to develop these skills in g education. The motive given e is thatthe e l will need all of the skills y fo life-long , andself-evaluation must be one of these.

The n initiative d m a e fundamental motive, which was tosee self- and t as an c t of any e which aims tohelp students to take e y fo thei own . A t slant onthis aspect is d in the study by James of d dialogues between sand students (1990). This study showed that in such dialogues, the s poweeasily s the student's , the latte being too modestly tentative.The effect is that y into the s fo a student's difficulty is not .Some of the h discussed in the section on m e aboveinvolved s e k on goals was d both with and without

g in self-evaluation; an example is the h by Schunk (1996) whichshowed that, if combined with e goals, self-evaluation e d

, self-efficacy and achievement.Some s have taken the t by developing a l

n on how students might change thei . The assumption eis they cannot do so unless they can d the goals which they e failingto attain, develop at the same time an w in which they can locate thei ownposition in n to those goals, and then d to e and e

g which changes thei g , 1989). n this view, self-assess-ment is a sine qua non fo effective . This l stance will be

d at the end of this section and in the section titled s fo the yand e of e assessment.

Studies of Self-assessment

h studies of self- and t can be y divided into twoe involving l k yielding quantitative data on achieve-

ment and those fo which the evidence is qualitative. These will now be discussedin . Two quantitative examples have y been d in some detail in thesection on m e (Fontana & , 1994; n &White, 1997). h of these have in common an emphasis on the need fo studentsto d the g goals, to d the assessment , and to have

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 27

the y to t on thei . evaluation played a t only in then & White study.

Two studies have d with n who have g difficulties. n theof these y & , 1992), the l g s of y schoolstudents e d by giving them l and visual e feedback,eithe by the teache only, o h , o . The

t gains, d by n of - and post-test s ove the -s d of nine weeks, e achieved by the g , whilst

all e did bette than a l p who had no e feedback. h on thes of acceptability to the s involved and on the y of thei ownl of thei , the - and g methods e d and

one benefit of both was that they d the amount of time that the specialeducation s had to spend on t in thei . n the second

h (Sawye et ah, 1992) the focus was on the g composition skills of 4thand 5th e students. , a p who e taught d s withexplicit attention to goals did bette than a simila p without the goal emphasisand a p without g . The p e bette lon n of the g skills taught, but all s with feedback did ,afte the t was , than othe g disability studentswithout any e of such feedback.

n h to investigate the most effective way of using a ge e s & , 1991), two s of 5th and 6th e

students e both given g in thei e use of the , but oneof them also had to take t in g , d by the s asmeta-cognitive . e was also a matched l p who used the

e without the . The g s e d by abooklet of questions with which students d thei s on a set of e

g s selected m the . h d s achieved success with the e than the l , but those with the

g g e also significantly bette than those without it. They ee successful with the e complex , they succeeded e quickly, and

l they e seen to be employing e effective . They seemed to do, not because they could use the s e effectively, but

because they d by g on a m and g the possibilities ofusing t s e n outcome which seemed to link withthe meta-cognitive emphasis g the g .

A focus on d g was seen, in the w by Thomas (1993), to bea y concomitant to the moves to develop l , study skills, and

y fo g amongst students. e distinguished e s thate independent , such as test w handouts, those thate it, including extensive e feedback, and d evidence

which established that such activities can e student achievement. n a wof the e of , n & g (1997) discussed the t

s of the e of n employed by l well-known s andlinked this to h evidence on the effectiveness of g students by

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 28 Black & D. Wiliam

g g (Schunk & , 1993b; n & ,1994). A closely d set of studies by g (1994) on students' questioning

s will be d in the section on Questions below.Self-evaluation is an c aspect of n on one's own . l

qualitative studies t on innovations designed to foste such . nscience education, d et al. (1991) d on k with 27 s and 350students e s e helped to know e about thei students and to n

e about how they might change the style of m k by a y basedon meta-cognition and . h the s and the students involvedhad to analyse what had happened in a piece of the g , and each side hadto e e changes to be put into effect. , students had to evaluatewhethe these changes had happened. The evidence, based on s by thoseinvolved, was that successful implementations had been achieved. d & i(1991) d a class of high-school students in g of thei tests and foundthat thei e gains e significantly highe than those of a l p class:they d this to the g of thei students' l t of andantagonism s d feedback. Simila success was achieved by t &

t (1992) in an t aimed to help students to , h feedbackon thei self-assessment, the lack of e between thei n ofthei k and the judgments of ; the quality and depth of the students'self-assessments e enhanced as the t . Simila k is

d by s & s (1993), l & n (1994) and &f (1997).

A scale innovation is fully d in a book by s et al. (1993). Theaim was to change assessment of achievement in the visual s by g studentsinto the assessment s as e , mainly h the develop-ment of 'assessment ' in which students e d to t onthei k and to e thei . The s e enthusiastic in theiaccounts of the success of thei , and believe that the students involved showedthat they e capable of and sophisticated s to and s ofthei own k ... in n with thei n ' (p. 161). Theyconcluded that the h opened up new s in aesthetic knowing and

, but that it also d that s abandon l assessment. , the evidence of the 'success' of the k is to be found only in

the accounts, d with quotations, of the quality of the students' aestheticjudgments. y qualitative s e given of an initiative to hand ove all

y fo assessment of a e e to students' self-assessment s & Sutton, 1991), and of the outcome of a t to n2nd, , and 4th e students to d thei on o off task state of k at

s (Wheldall & , 1992). n both cases,the initiative d a significant change in students' commitment to thei kand e was also some t evidence in both of t in thei gachievement.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 29

l of the accounts d in this section involve both self-assessment and. t as such is included in l accounts of the

development of p n as a t of m g activity. n anl study by h & Shulamith (1991), college students e taught to

e thei own questions about topics in physics, and achieved bette ggains than those who used only s questions; amongst those g theiown questions, some also used pee feedback to answe and discuss thei , andthis p showed even g gains than the . s et al. (1994)also used e , in thei k with 1st and 2nd e ndeveloping assessment skills in thei d t . The n dthei own , and the quality of these e g the study. Good twith ' assessments was achieved, with n tending to .

, s e not e in thei assessments of othe . Thes of self- and s e also investigated, in k with college

biology students, by Stefani (1994). e found s with ' assess-ments of 0.71 fo self-assessments and 0.89 fo . All of the studentssaid that the self- and t k made them think , and 85%said that it made them n . s & e (1993) also investigated

t of final yea s in y and found a -lation coefficient of 0.83 between the mean s of s and those of a p ofstaff.

t is often difficult to disentangle the t activity othe novelactivities in k of this kind, and impossible in l to e any d gainsto the assessment component. l s e given by Slavin (1991) and byWebb (1995). The second of these does focus on assessment s in p

k and it s the e of g in p s and of the settingof clea goals and clea achievement . n such , a clea choice has tobe made, and d in the , between a goal of the best e m the

p as a , and a goal of g individuals' s h p. The question of the optimum p composition is a complex one;

e a p goal has , then fo well defined tasks, established highs e the most , but fo e open tasks a e of types of

students is an advantage. e individuals' e has , then the highs e little affected by the mix, but the low s benefit e m a

mixed p d that the p g emphasises methods fo g out, than , thei . The need fo such e is emphasised

in a study of p discussions in science education by Solomon (1991).

Links to Theories of Learning

The s given by Zessoules & (1991) show how any assessmentchanges of the types d above might be expected to enhance g if theyhelp students to develop e habits of mind. They e that such

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008

30 Black & D. Wiliam

development should be an essential component in s fo the implemen-tation of authentic assessment in m . Assessment is to be seen as amoment of , and students have to be active in thei own assessment and to

e thei own g in the light of an g of what it means to get.

n , it can be seen that these s s to developing self-assessment by pupils hold e of success. , thei n in

n to e l s of g s fundamental , asd by the analysis of Tittle (1994). Full discussion of this and simila k

will be d until the last t of this . A few points can be d. n a w of n h in this field, (1994) points out

both that students e often unwilling to give up y need tobe convinced h discussion which s thei own n on theithinking—and also that if a student cannot plan and y out systematic l

g k fo himself, he o she will not be able to make use of good efeedback. h of these indicate that self-assessment is essential. , e etal. (1996) e that t teaching of study skills to students without attention to

, meta-cognitive, development may well be pointless. One n fo theneed to look fo l change is that students g to thei k models of

g which may be an obstacle to thei own . That pupils do have suchmodels that e to a e y d is d by a n ofthe s to g of n and Japanese students e & ,1996), whilst the finding that the most able students in eithe y e e alikethan thei s in having developed simila effective habits of g shows thatsuch g s can be .

The task of developing students' self-assessment capabilities may be das a task of g them with e models of this way of . na modest way, l (1994) d to do this by g d examples of

a s to students fo them to study, g some of the kon solving s fo themselves which they would y be doing. Theachievement of these students was d by this method, and the low sshowed y good . The autho d that fo manystudents, the task of tackling new s in a new a of k might notbe useful because of cognitive . Study of a d example d a lessloaded g situation in which n on the s used could bedeveloped. e , s discussion (1991) leads to the conclusionthat the teache must e a model of g fo the student, andneeds also to be able to d the model in the head of the so thathe/she can help the to g into his o he 'meta-cognitive haze'. Thedifficulty e is that many s do not have a good model of gand of effective g to , and e lack both the l

k within which to t the evidence d by students and themodel to which to t them in the development of thei own self-assessment

.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008

Assessment and Classroom Learning 31

Strategies and Tactics for TeachersOverview

The s aspects of a s k in e assessment can be d inn to the l sequence of decisions and actions that e entailed. This

h will be used e as a fo g the k d in the. Thus, the sub-sections below will deal in n with the choice of tasks,

with m , with l aspects of the use of questions, with tests andthen with feedback tests. A closing section will then look at s ,including k that looks e deeply at the assumptions and s that might

e the n of tactics.

Choice of Task

t is obvious that e assessment which guides s s valuedg goals can only be d with tasks that both k to those goals and

that e open in thei e to the n and display of t evidence,both student to teache and to students themselves. n thei detailed qualitativestudy of the m s of two outstandingly successful high-schoolscience , t & Tobin (1989) concluded that the key to thei successwas the way they e able to monito fo . A common e was the

y of class activities—with an emphasis on questioning in which 60%of the questions e asked by the students. n a e l w of m

, Ames (1992) selects e main s which e success-ful ' (as opposed to e section on Goal n above)

. The t of these is the e of the tasks set, which should be noveland d in , offe e challenge, help students develop m

d goals, focus on meaningful aspects of g and t thedevelopment and use of effective g . d (1992) ssome of these issues, pointing out that such notions as 'challenging' and 'meaning-ful' e . A task e the challenge goes too fa can lead to studentavoidance of the involved, and fo students who e fa behind it is difficult to

e thei s without at the same time making them e of howfa behind they . , tasks can be meaningful fo a y of sand it is t to emphasise those meanings which might be e fo

.n an w about science teaching, e & (1987) ee ambitious. They emphasised the need to shift t pedagogy to give e

emphasis to l aspects of knowledge and less to the e aspects.They outlined a scheme fo the e analysis of tasks which could bedeployed by s to e a e analysis of the tasks they e using.This scheme distinguished tasks which (a) d a specific situation identical tothe one studied, o (b) d a 'typical' m but not one identical to the onestudied, g identification of the e m and its use, thanexact n of an e as in (a), and (c) a quite new m

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008

32 Black & D. Wiliam

g new g and n of a new , deploying establishedknowledge in a new way. Students would need special and explicit g fotackling tasks of type (c). They d that all e types of task e needed,but that s do not y plan o analyse the tasks that they set by anyscheme of this type. Such t is an essential condition fo planning the

n of e assessment, both fo n of feedback and foplanning how to d to it.

Discourse

That the quality of the e between teache and students can be analysedat l t levels is evident the extensive e on m

s and e analysis. n a w of questioning in ,n (1991) s the t h with the socio-linguistic

, and s that inconsistency of h s on the cognitivelevel of questioning may be due to neglect of the fact that the meaning of aquestion cannot be d m its e s alone. As both he andFile (1995) , the meaning behind any e depends also on thecontext, on the ways in which questions in a m have cometo signify the s of s between those involved that have beenbuilt up ove time. & e (1996) give an example of how a habitual

n can be unhelpful fo . Newmann (1992) makes a plea, onsimila , fo assessment in social studies to focus on , definedby him as language d by the student with the intention of giving

, , explanation o analysis. The plea is based on an tthat t methods, in which students e d to use the languageof , e the e use of e and so esocial knowledge. A g example of such effects is d in a pape byFile (1993). n the same vein, Quicke & Winte (1994) t success in kwith low achieving students in Yea 8, e they aimed to develop a social

fo dialogue about . The k of s et al. (1993) inaesthetic assessment in the s can be seen as a e to this plea, and thedifficulties d by (1994) e evidence of the inadequacy of established

.s pape uses the e 'qualitative e assessment' in its title, and

this may help to explain why quantitative evidence fo the g effects ofe s is d to find. An exception is the h by e (1988)

on m dialogue in science . e analysed the e of es in fou , g the quality of the e by summation ove

fou . These included the s of e themes, the s ofs (an indicato of ) and s of themes explicitly

d to the content of the lessons. This e e was included with eothe , of scholastic aptitude, locus of l and n level -ively, as independent , and an achievement post-test as dependent .With the class as the unit of analysis, the e e accounted fo 63% of

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 33

the , with the e s accounting y fo unde 4%, 22% and14%.

Johnson & Johnson (1990) t a meta-analysis to show that ee can e significant gains in . s & l (1995),e & n (1996) and l & Gitome (1997) all t k with

science s to e such . The n in all e cases was tohelp students to move, in talk about thei , focus in y andcontent based s s a deepe discussion of conceptual . h &

y (1994) d the use of concept maps as an aid in suchdiscussions; such maps, n by the students, e to e useful points of

e in g the points unde discussion and enable the teache to engagein 'dynamic assessment'.

Questions

Some of the t aspects of questioning by s have y been -duced above. The quality of m questioning is a matte fo , as

d in the k of Stiggins et al. (1989) who studied 36 s ove a eof subjects and ove s 2 to 12, by n of m , study of theidocumentation, and . At all levels the questioning was dominated by lquestions, and whilst those d to teach thinking skills asked e

t questions, thei use of questions was still . Anexample of the l t was that in science , 65% of the questions

e fo , with only 17% on l and deductive . The sfo n k e simila to those fo the l . e & g (1994)studied the m s of novice s in mathematics and found thatthey tended to t students' questions as being individual , sthe s of t s tended to be d e to a 'collective student'.

l s t k focused on question n by students, and aspointed out in the section on Self-assessment above, this may be seen as anextension of k on students' self-assessment. With college students, g (1990,1992a,b; 1994) found that g which d students to e specific

g questions and then attempt to answe them is e effective thang in othe study techniques, which she s in s of the y

g the g which aimed to develop autonomy and 'l ove thei own . Simila , which also showed that students' own

questions d bette s than adjunct questions the , ed fo i college students by h & Shulamith (1991). n k with 5th

e school students, a simila h was used in g students withg on compute d tasks , 1991). With a sample of 46

students, one p was given no a , anothe was d to ask andanswe questions with student , whilst a d p e also d inquestioning one anothe in s but d to use c questions fo guidancein cognitive and meta-cognitive activity. The latte g focused on the use of

c questions' such as w e X and Y alike?' and 'What would happen if...?'.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 34 . Black & D. Wiliam

The outcome was d by a post-test of n s and a novel computetask. The p d to ask c questions of one anothe d the

. Foos et al. (1994) have d simila success on outcome s whenstudents e d to e fo examinations by l techniques, the mostsuccessful being the n of thei own study questions followed by attempts toanswe them. This , and simila k in school science classes g &

, 1993) can be seen as t of a y to e dl thinking , 1995).

Such k has two main elements. One is the n of thinkingand n of thei study by students h the n of questions,the othe is to conduct this development h pee . A e

w of studies of this type by e and colleagues (1996) s ameta-analysis of selected studies. The effects e y positive, but the effect sizedepends on whethe the outcome e is a d test, o a ntest developed by the . The latte give effects, with means of 1.00fo 5 studies with l pee questioning and 0.88 fo 11 s without this

e (the e between these was not significant). The conclusion is thate is no evidence that pee n is to t n in question

. t is pointed out that the l e fo active g bystudents does not e specific guidance about the choice of methods, and the

w discusses the e s adopted in some detail.A t use of questioning is to e and develop students'

knowledge. A w of k of this type y et al., 1992) establishes thatg s to compose s with explanations to e thei

knowledge of new k does e , and that this may be because it helpsthe to e the new to the old and to avoid l judgments about thenew content.

Anothe y of questioning is the use of adjunct questions with text. eis little to add e to the studies d by . A study by y & n(1991) with a high school biology class showed that when the teache d kon questions of the type to be used in a test and emphasised thei ,

e was . Anothe study with the use of n adjunctquestions was aimed to e concept g with science lessons in which

d sequences e used y & , 1992). Eighthe students e assigned to a l p o to one of fou t .

The s d that the questions used did succeed in thei aim of focusingstudents' attention on the concepts involved, and that e they e used fo onlythe of the 12 sequences used, they d effects on the way in which those

g e studied. The s composed the questions to e a l aimof scaffolding the meta-cognitive activity of the students. , such kshould be d in the light of the s quoted in the section on Choice of taskabove (and o & , 1990; , 1994) which indicated that it mighthelp to employ some of the g time of students on othe l tasks in placeof the time spent tackling questions.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 35

The Use of Tests

One study giving evidence that testing can lead to d g hasy been cited in the section on m e above z &

, 1992). , et al. (1991b) d the evidence of theeffects of class testing. Thei meta-analysis of 40 t studies showedthat e d with testing and d with d

up to a n level, but that beyond that e beyond 1 and 2tests pe week) it could decline again. The evidence also indicated that l ttests e e effective than fewe longe ones. Simila evidence is quoted by

(1991, 1992, see the section on Task motivation , below).n a late investigation with college students in psychology, n et al. (1994)

found that the addition of d tests d no significant -ment in , even though the students in the t said that theywould like to have such tests in othe s also. A y negative t was

d by z (1989), but a , positive effect was found by Schloss etal. (1990) g with e students in teache g fo special education.When given a t e quiz afte each e students dsignificantly bette than they did when no quiz was d on e —post-tests of familia items, post-tests of unfamilia items and a y of satisfactionwith the .

n some subject , s e t to use tests fo fea of inhibiting. t (1996) attempted to tackle this m with y school

s in the assessment of . The t d with both d ande s to develop a and a language fo assessing , and

this the p e able to e guidance about the feedback e ton g to the classification in the to which thei k was

judged to . s of new assessment methods e tospecific subjects e also d by Adelman et al. (1990) fo visual and -ance s with olde y students and by t (1993) fo y in yschools.

The aim and e of the tests employed e not explained in most studies.s speculated that low level aims might benefit e testing, but

that highe level aims would benefit lowe f & a (1992)conducted a e g w and d with s on this point. Theyselected 20 studies, of which 18 gave positive effects. They point out that the eof the n test could t . n any such study, the n test mightbe e t to a l p than to the t . On the othehand, if the final summative test e to contain questions simila to those in theclass tests, e could be a n in the e . They concluded thatonly fou of the studies they d e this type of flaw. The sof these e all positive with a mean effect size of 0.37, but they e all withcollege students. Following thei l , these s e the s ofan investigation with 2000 students in 93, 10th , classes in Saudi .

l classes e given the l monthly quizzes, whilst the s e given

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 36 Black & D. Wiliam

bi-monthly quizzes. The a e tests set by an investigato who had not seenany of the quizzes d and used by the , and included a test at theend of the e and a delayed test e months . t and lclasses e set up in s having the same teache fo the two s of each

. e e significant s between the two s on both testoccasions, with an effect size of about 0.3 in favou of the e y tested

. The effect was much fo high s than fo the medium and low. , the tests used e composed only of e o multiple

choice items, so that the g at issue was y l in . Thee taken with this t shows how studies cannot be accepted as t

without l y of the t design, and of the quality of the questionsused fo both the t and the n tests.

d these s lies the issue of whethe o not the testing isg the e assessment function. This cannot be d without a study

of how test s e d by the students. f the tests e not used to givefeedback about , and if they e no e than s of a final high-stakes summative test, o if they e components of a continuous assessment schemeso that they all bea a high-stakes implication, then the situation can amount to no

e than t summative testing. Tan (1992) s a situation in a efo yea medical students in which he collected evidence that the tsummative tests e having a d negative influence on thei . Thetests called only fo low-level skills and had y established a 'hidden -lum' which inhibited high level conceptual development and which meant thatstudents e not being taught to apply y to . Simila s aboutthe cognitive level of testing k is d by l et al. (1995, discussedin the section titled Expectations and the social setting, below).

The Quality of Feedback

h of the s sub-sections lead to the almost obvious point that the qualityof the feedback d is a key e in any e fo e assessment.The l effect of feedback m tests was d by s etal. (1991a) using a meta-analysis of 58 s taken 40 . Effectsof feedback e d if students had access to the s e the feedbackwas conveyed. When this effect had been allowed , it was then the quality of thefeedback which was the t influence on . d nand simple completion assessment items e associated with the smallest effects.Feedback was most effective when it was designed to stimulate n of s

h a thoughtful h to them in n to the l g tto the task.

The feedback d by ' n s to students' k wasstudied in an t with ove 500 Venezuelan students involving 18 math-ematics s in e schools (Elawa & , 1985). They d the sto give n feedback which d on specific s and on poo ,with suggestions about how to , the whole being guided by a focus on deep

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 37

than l . A l p followed the l e ofg the k without comments. n to check whethe effects of the

feedback g on thei teaching could account fo any , a d p ofthe d s d half of thei classes with full feedback and the othe halfwith s only. All e given a t and one of e l s ofpost-test. Analysis of e of the s showed a big effect associated with thefeedback , which accounted fo 24% of the e in final achievement(with anothe 24% associated with achievement). The t also dthe initial y of boys ove s and had a e positive effect on attitudes

s mathematics.n a quite t e of , Tenenbaum & g (1989) d out

a meta-analysis with 16 studies of the effects of 'enhanced , involvingemphasis on cues, , , feedback and , on motoskill g in physical education. e to these s of enhancement

d gains with a mean effect size of 0.66, and they also enhanced students'time on task.

The linkage of feedback to assumptions about the e of the student gwhich it is designed to e has been taken in k on -based assessment by Fuchs et al. (1991). Thei t with mathematicsstudents d the possible t of a scheme of systematic assessment ofstudent development by setting up an t system' which s could consultto guide thei l planning in n to the students' assessment .The t used e s of , one who used no systematic assess-ment, a second p who used such assessment, and a d who used the sameassessment togethe with the t system. h of the second and d s

d thei teaching s e y than the , only thed p d bette student achievement than the and whilst s

in the second p d to feedback by using t s withoutchanging the teaching , those in the d d both. The conclusion

d was that s need e than good assessment y alsoneed help to develop methods to t and d to the s in a eway. One t fo such an h is a sound model of students' -

n in the g of the subject , so that the a that guide thee y can be matched to students' s in ; this need,

and some evidence g on ways to meet it, has been studied fo both schoolmathematics and school science , 1993a, pp. 58-61; s & Evans, 1986;

n & , 1987). Fo such data, the n sequences have to be attunedby e data to indicate e expectations fo students at t ages.This has been attempted with data fo spelling, g and mathematics by Fuchset al. (1993), who conside both the l needs and the implications of suchdata fo developmental studies of academic .

A e e discussion of feedback will be d in a section belowdevoted to this topic.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 38 Black & D. Wiliam

Formulation of Strategy

The sub-sections above can be d as s of the s components ofa kit of s that can be assembled to compose a complete . The hstudies d can be judged valuable, in that they e a complex situation by

g one e at a time, o as flawed because any one tactic will y in itseffect with the holistic context within which it . t has also d that atleast some of the accounts e incomplete in that the quality of the s o

s that evoke the feedback, and of the assumptions which m then of that feedback, cannot be judged. At the same time it is clea that

the fundamental assumptions about g on which the , sand s e based e all .

l s have n about the c , and e hasy been made to some of thei . The analysis of Thomas et al. (1993)

stands out because they have attempted a quantitative study to encompass the manys involved. They applied l linea modelling to data collected m

12 high school biology , focusing on the s of the s which placeddemands on and gave the t to the students. At the student level, the sindicated a positive link between achievement and both thei self-concept of aca-demic ability and thei study activities; these last two e also linked with one

. Students' engagement in active study k was positively associated withthe n of challenging activities and with extensive feedback on thei k inthe . Such feedback was also y d with high achievement. Amongstthe othe s that e teased out was the finding that twhich d e demands d the p between self-conceptand achievement. This k constitutes an ambitious , but, almost of necess-ity, only y l n is d about the g quality of the

k being studied.Weston et al. (1995) have d that if the e on e assessment is

to m l design, then a common language is needed. They identifyfou components—who , what s can be taken, what techniques canbe used and in what situations these can , and e that l designshould be based on explicit decisions about these , to be taken in the light of thegoals of the . The model was used to analyse 11 l tests and

d that e e many assumptions about these fou issues which eembedded in the language about e evaluation.

h Ames and Nichols attempt e ambitiously detailed analyses.Fo Ames (1992), the distinction between the e and y sis a g point, but she then outlines e salient , namely meaning-ful tasks, the n of the ' independence by giving y tothei own decision making, and evaluation which focuses on individual -ment and . The e of changing the assumptions that smake about g is d in this . The analysis s many

s to that of Zessoules & (1991). An account of a t toe and t s in making changes of this type , 1989) s

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 39

out the many difficulties that s , both in making theiassessments g to g , and in changing thei teaching and feed-back to k away m d assumptions in g student's

.Nichols's (1994) analysis goes deepe in g on what he s cogni-

tively diagnostic assessments. t is pointed out that classical s has beend s the use of assessments to guide selection, and so a new p

with cognitive science is needed if it is to be used to guide . Tests must bedesigned in the light of models of specific knowledge s in to help

e the s of s in g those , so that the in-n of the feedback can e the e of making s about

students' cognitive mechanisms. t is clea that many l types of test einadequate to this e because they do not l the methods used by thosetested. h et al. (1992) d the s which affect the validity ofassessment tasks when judged t e and emphasised thata majo t to validity is the extent to which students can t the meaningsof the tasks intended by those who set them. h in thei account and that of

e & (1995) the analysis is d by detailed s of the kof one o two . , the latte , developing the s in

e (1993), e a e g l discussion, gtwo s to e assessment—a t one, g tagainst objectives, and a social t one g the assessment into

. , fo the assessment of language, Shohamy (1995) s that thecomplexity of language calls fo a special discipline fo language assessment,

d in a clea l e of what it means to 'know a language'.Thus task selection, and the type of feedback that a task might , e

a cognitive y which can m the link between ' g andthei s with assessment tasks, in the light of which assessment activitiescan be designed and . Such an h will of e t ywith the pedagogy adopted, and may have to tempe an a i l positionwith a s to adapt and develop by an inductive h as efeedback challenges the e of the k (Fuchs & Fuchs, 1986). The con-clusion of the analysis is that a y substantial , involving nbetween , cognitive scientists and subject s is needed.

All of these discussions point to the need fo y g changes if eevaluation is to e its potential. Some e changes in pedagogy haveattempted to meet such , and e distinguished what has been discussedin this section by thei e and c . These will be thesubject of the next section.

Systems

General Strategies

Good assessment feedback is eithe explicitly mentioned o y implied in

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 40 Black & D. Wiliam

s of a e of studies and initiatives in which such feedback is one componentof a . Thus, fo example, in g a study of school effective-ness e et al. (1988) point out that feedback and good g ekey aspects of effectiveness. n the initiative in n to develop a holistic dof Achievement' to cove all aspects of a student's k and n within aschool, the student is to be involved in negotiating an d . Thus self-assessment is d to be an t , but it has been d in a

of ways and is sometimes l , 1992; t et al.,1990). Enhanced attention to diagnosis and n is a e of many otheschemes, fo example g y and Slavin's Success fo All scheme (Slavinet al, 1992, 1996).

Assessment and feedback e also an t e of y g, discussed in e detail below, but with many (if not all) of these

teaching systems, even identifying the e e of the e feedback used,let alone its n to the global s in attainment , isdifficult. Fo this , these systems e d only y in what follows.

Studies of Learning

y g d as a l implementation of the g s ofJohn . . e d that success in g was a function solely of the

o of the time actually spent g to the time needed fo n othe, any student could n anything if they studied it long enough. The time

spent g depended both on the time allowed fo g and the se while the time needed to n depended on the s aptitude, the

quality of the teaching, and the s ability to d this teaching (seek & , 1976, p. 6). Two main s to y g e

developed in the 1960s. One, developed by n , using dd teaching , was called g fo y ) and the

othe was s individual-based, student-paced d System of -tion . The vast y of the h n into y g hasbeen d on , than s model, and that which has been doneon is y confined to and highe education.

A key consequence of s LF model is that students of g aptitudewill diffe in thei achievements unless those with less aptitude e given eithe a

y to n o bette quality teaching. Fo most s ofy , this is not to be achieved by g teaching s at

students of lowe aptitude, but by g the quality of teaching fo all students,the g assumption being that students with highe aptitude e bette ableto make sense of incomplete o poo n t & , 1989).

The key elements in this , g to l (1969) : The must d the e of the task to be d and the

e to be followed in g it. The specific l objectives g to the g task must be

.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 41

t is useful to k a e o subject into small units of g and to test atthe end of each unit.

The teache should e feedback about each s s anddifficulties afte each test.

The teache must find ways to alte the time some students have available to .t may be e to e e g .

Student t is d when small s of two o e students meety fo as long as an hou to w thei test s and to help one anothee the difficulties identified by means of the test.

, although the s of y g cove all aspects of gand teaching, effective e assessment is a key component of effective y

. e majo s of the h into the effectiveness of yg use the technique of 'meta-analysis' to combine the s y

of t studies. The w of k & s (1976) s k done in the half of the 1970s while Guskey & Gates (1986) and k et al. (1990) cove

the subsequent decade.n them, the s by k & s (1976) and Guskey & Gates (1986)

e 83 s m 35 studies) of the effect of y g on lachievement, all using the g Fo ' h . They found an

e effect size of 0.82, which is equivalent to g the achievement of an' student to that of the top 20%, and one of the t e effects eve

d fo a teaching y k & , 1989). When the age of the studentsinvolved is examined, it s as if y g is less effective fo oldestudents. it is not clea whethe this is because olde students e e 'setin thei ways' and e have e difficulty in changing thei ways of gto those d fo y , o because y g is adapted e

y within y school a and pedagogy.These s have also d whethe y g is e effective in

some subjects than . k & s (1976) found that s fo science (andg to some , sociology) e e consistent, but lowe than fo

othe subjects, while s fo mathematics , on , , but fa lessconsistent. , while Guskey & Gates (1986) also found low effect sizes foscience, these e e to the effect sizes fo mathematics, and both esubstantially lowe than those found fo language s and social studies. Thus noclea consensus s h on the e effectiveness of y

g s in t subjects.The 1990 w by k et al. looked at 108 studies which e judged to meet

thei a fo inclusion in a meta-analysis. Of these, 91 e d out withstudents ove 18 s of age, 72 using s h and 19 using sLF . The 17 school-based studies all used , although these ealso skewed s olde students—only two of the studies contained any s

students younge than 11 s of age. The effect sizes found e smalle thanthose found by k & s and Guskey & Gates—not g given the

n of studies with olde students. , k et al. also

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 42 Black & D. Wiliam

found that the self-paced h tended to have smalle effect sizes than thed LF , and also seems to e completion s in college

. The e s also found that g s e eeffective fo g students, thus tending (as y intended) to ethe e of achievement in the , although , such as Livingstone &Gentile (1996), have found no evidence to t the g yhypothesis'.

l to the issue of whethe y g is e effective fo -achieving students is that of whethe it is equally effective fo all . z& z (1992) looked at the effect of d testing in a l -uate mathematics e and found that t ' testing was effective in

g achievement, but it was e effective fo the less d . don a meta-analysis of 40 studies, t & s et al. (1991b) estimated thattesting once y e weeks showed an effect-size of 0.5 ove no testing,

g to d 0.6 fo weekly tests and 0.75 fo twice-weekly tests., most notably t Slavin, have questioned whethe y g is

effective at all. n his own w of h on y , he smeta-analysis as being too , because of the way that s all the

h studies that satisfy the inclusion a e . s own h isto use a 'best-evidence' synthesis (Slavin, 1987), attaching e y subjec-tive) weight to the studies that e well-designed and conducted. Although many ofhis findings e with the meta-analytic s d above, he points out thatalmost all of the e effect sizes have been found on , than

, tests, and indeed, the effect sizes fo y g d byd tests e close to . This suggests that the effectiveness of y

g might depend on the ' of the outcome .This is d by k et al.'s (1990) finding that the effect sizes fo y

g as d by e tests (typically d 1.17) e than fosummative tests d 0.6).

Slavin (1987) s that this is because, in y g studies eoutcome is d using d tests, the s focus y onthe content that will be tested. n othe , the effects e d (eitheconsciously o unconsciously) by 'teaching to the test'. The x of this -ment is e the e of ' of a domain—should it be the -

d test o the d test?

The of Learning

The only clea messages g the y g e e thaty g s to be effective in g students' s on -

d tests, is e effective in d s than in self-paced, and is e effective fo younge students.

, while establishing that unde n , y gis effective in g achievement, the e gives y little evidence as towhich aspects of y g s e effective. Fo example, while

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 43

most of the h s on the effects of y g on students, theexplanation of the effects could be that g fo teaching fo y s

l development fo the teache (Whiting et al., 1995)., one of the s voiced by the s cited above is that too often

it is impossible to establish the h s which s of yg e implemented in the h being , let alone which e

effective (Guskey & , 1988). e e at least five aspects of 'typical' yg s that e t to the s of the t :

that students e given feedback; that students e given feedback on thei t achievement against some

expected level of achievement (i.e. the ' level); that such feedback is given ; that such feedback is (o is at least intended to be) diagnostic; that students e given the y to discuss with thei s how to y

any weaknesses.

, it is by no means clea that all of these e y to achieving the gainsclaimed fo y . Fo example, k & k (1987) e that ytesting is a l component in the success of y , and that itsomission leads to a substantial p in a s effectiveness. On the othehand, s without d ' testing, but with many of the othe

s of ' s can show significant effects.Fo example, in a study by n et al. (1996) a p of low-achieving second

e students e given a yea of l s ' , inwhich s explained and modelled , gave additional coaching asneeded, and d students to explain to each othe how they used the

. At the end of the , this p d by a e na simila p taught by highly d s using e l methods(exact effect sizes e not given, but they e between 1 and 2 d devia-tions).

Assessment Driven

An account has y been given in the section on m e of then of a complete system of planning and t fo n

, in which the wholesale innovation s to have had e assess-ment as a l component, so that it seems valid to e the d successto that component n et al., 1991). Anothe wholesale h is to m

e h application of the concept of scaffolding y & , 1993;n & , 1997). t again e s e the m of

acting on assessment s is tackled by g the k on amodule o topic in such a way that the basic ideas have been d by about

s of the way h the ; assessment evidence is d at thisstage, so that in the g time d k can d g to theneeds of t pupils k & , 1984; , 1988).

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 44 Black & D. Wiliam

Two e loosely defined systems e those d as dassessment' and those d as ' systems.

d assessment ) is a development which expanded in the late1980s. The focus of many of the studies has been on y s education and theidentification of pupils with special educational needs, but its methods and -ples could apply s the m of education. Useful s of the

e e to be found in the collection edited by (1993), and one ein that collection (Shinn & Good , 1993) sets out the l s of A asfollows:

Assessment s should faithfully t the main g aims and shouldbe designed to evoke evidence about g needs.

The main e fo assessment is the e . Validity is n as g that l decisions taken on the

basis of assessment evidence e justified. The focus of attention is the individual and individually attuned l

action. The n m assessment should e to locate the individual's attain-

ment in n to a fo , but that this location should also bed by m data on the s of s g to the same .

The assessment should be so that the y of g ove time canbe : the t of g success is the key o follow eachpupil's s in , and to indicate cases of special need.

Shinn does not make a p distinction between A and the d concept ofd t , but s insist on this distinction (Salvia

& , 1990; Salvia & Ysseldyke, 1991; , 1993; Tindal, 1993). , foexample, sees as a sub-set of A d with specific s and

s focused on basic skills fo the diagnostic k of special education, and also s the title's m ' as g the -

ance in the y of g s to an established quantitative scale. eis also some t as to whethe o not A can be d as a

' .A n seen as t by two s is to distinguish A

y g , 1993; Fuchs, 1993). They see y g as gthat s follow a specific skill sequence step by step, which s g tofollow a path, s A is fa and loose and so allows fo

s to follow a y of s to , with less emphasis on g eskills in isolation.

The h evidence about is d by Fuchs (1993). Fo , the settingof explicit g goals is a distinctive e of . The h evidence is thatstudents achieve highe levels of attainment if the g goals e ambitious fo them.

s have also d those g to static goals, set at the outset andnot subsequently amended, with those g to dynamic goals, which e amended,usually d with g changes in the , in the light of

d . The dynamic h leads to bette achievements.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 45

The n by Shinn of what he calls the m of A makes it clea thatthis is a e assessment , and that many of its s would be essentialin any n of e assessment into a g . What maybe distinctive is the insistence on y focused test designs, and on the use of

t tests to give s of e against time as a key diagnostic.

The o movement is e closely associated with s to change the impactof high-stakes, often , testing of school . e is a vast eassociated with the o movement in the USA. h of it is , by Collins(1992), in the edited collections of f & n (1991) and—fo assessmentof y Calfee & o (1996a), whilst s & y (1993), set outsome of the issues in highe education. s (1996) gives an account of the s ofthe innovation, g the k as an attempt in t to satisfy demands ofaccountability whilst avoiding the s of d tests.

A o is a collection of a student's , usually d by selection ma s and often d with a e piece n by the student tojustify the selection. The involvement of the student in g and selecting is seenas s s says g ways to e that kind of n on a widescale has been at the t of the t assessment m the beginning' , 1996,p. 192) and, speaking of the e of students 'What was g was thei abilityto t on thei own k in n to a set of d sthat they d with many ' , 1996, p. 194). , g about a

, national, , o (1996) s on the ' enthusiasm, bothfo the powe of s to focus student attention on thei own g s andaccomplishments, and fo the evidence that s believe the k changes the waysin which they teach and s thei expectations fo thei students. Calfee &

n (1996) see s as g a technology fo helping the slogan ofd ' to become a . s n et al., 1996) emphasise

that it is valuable fo students to d the assessment a fo themselves,whilst Yancey (1996), in a e subtle analysis of the concept of n as ,points out that the e of helping students to t on thei k has made

s e e fo themselves., e is little by way of h evidence, that goes beyond the s of

, to establish the g advantages. Attention has focused on the of ' g of s because of the motive to make them satisfy

s fo accountability, and so to e summative s as well as the. n this , the tension between the s plays out both in the

selection and in the g of tasks t & Yang, 1996). o (1996) sg s based on a multi-dimensional , with the n that

each dimension t an aspect of g which can be d by students andwhich s an t aspect of . , he identifies the m thus

t it does not y follow that it will be l to g national s

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 46 Black & D. Wiliam

into the focus of self-assessing students and thei ' , 1996, p. 241).Calfee & o (1996b) t on h with s into thei e of

using . The s showed a g gap between the l c andactual , fo they showed that many s e paying little attention to

l s and e g little evidence of any engagement of studentsin g why they e doing this . Thei conclusion was that the efate of the movement hung in the balance, eithe between e negative possibilities—

, , o becoming too the positive possibility ofg a d n in .

Slate et al. (1997) e an t in an y a e focollege students which d no significant e in achievement between a

p engaged in o n and a l . , the achievementtest was a 24-item multiple choice test, which might not have d some of theadvantages of the o , and at the same time the teache d thatthe o p ended up asking e questions about l d applications andhad been led to discuss e complex and g phenomena than the l

. n chapte 3 of thei book, s & y (1993) also t on an attemptto evaluate a g e with college students which also seemed to show only smallgains, but point out that they e assessing holistic g qualities which sassessments and g s of thei students had neglected.

Summative

The d Assessment schemes in England e e s designedto e l examinations fo public s by a s of dassessments, conducted in schools but d (i.e. checked fo consistency of

s between schools) by an examining . n that they d thel examinations by t tests within each school, and enhanced the

e of components of assessed k as s to the summative, they influenced the ways in which assessment was d within schools andd a distinctive o fo the g out of e tensions.

Whilst l accounts of these schemes have been published k & ,1986; Lock & , 1988; Swain, 1988, 1989; n & Lock, 1989; ,1990; Lock & Wheatley, 1990) e does not appea to be any published hwhich could identify the developments of the e functions withinthese schemes. A simila scheme in sciences, in that the summative function was linkedto t assessment ove extended s within m , has been

d by e (1992). n all of these accounts, one of the s that standsout is the difficulties that s and s met in g to establish a

d h to assessment. n l such schemes, an te has been the n of a l bank of assessment questions m which

s can w g to thei needs—but these have y beendesigned with summative needs in mind. n Canada, a et al. (1993) e thesetting up of a bank of diagnostic items d in a l scheme:diagnostic context, notional content and cognitive ability, the items being d

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 47

a study of common s so that they could e a basis fo causal diagnosis. Thel e was to help s to e e l feedback within the

s of l . s in five s showed that by nwith a l set of five e , those using the item bank had gains,the mean effect size being 0.7.

The s of developing d assessment beset the fa el s in Queensland , 1987; , 1995). This n state

abolished l examinations fo y schools in 1971, but subsequentlyd s in the quality and the g of school-based

assessments. s e s the development of a d, with s having to n the skills and the state having to develop

systems to e y of n of the n .emphasis on assessment d ove two s within m , on feedback tostudents on successive assessment , and on the n of student sas evidence fo the n , have been d in these developments.

, the impact of these on the e e of assessment has yet to be.

The systems d in this y should indeed have implications fo eassessment and could h the m of the e p

m a t n m most othe studies, e the high-stakes s eeithe , o accepted in that achievements on the existing s e used

) as the a of success. , e is little evidence that thee p has been thought out in thei design, and little

substantial evidence about how it has d out in e (but see e & , 1996and section titled e e implications fo policy? below).

Feedback

The two concepts of e assessment and of feedback . The mfeedback has d y in the account so , and the section on the qualityof feedback is explicitly d with the feedback function. , that sectionhad a limited focus, and the usages y have been e and not subject to

t consistency. e of its y in e assessment, it is tto e and y the concept. This will be done in this section as a y

e to the fulle w of e assessment in the subsequent final section.

The of Feedback

, feedback was used to e an t in l and cs y n about the level of an 'output' signal (specifically the gap

between the actual level of the output signal and some defined ' level) wasfed back into one of the system's inputs. e the effect of this was to e the gap,it was called negative feedback, and e the effect of the feedback was to ethe gap, it was called 'positive feedback'.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 48 Black & D. Wiliam

n applying this model to the l sciences, we can identify fou elementsmaking up the feedback system: data on the actual level of some e ; data on the e level of that ; a mechanism fo g the two levels, and g n about the gap

between the two levels; a mechanism by which the n can be used to alte the gap.

Fo & i (1996) only the t of these is y fo feedback to exist.They define 'feedback ' as 'actions taken by an l agent to e

n g some aspects of one's task , although it is hnoting that the t fo an l agent excludes . n ,

d (1983) defines feedback as follows:Feedback is n about the gap between the actual level and the

e level of a system which is used to alte the gap in someway (p. 4).

and specifically s that fo feedback to exist, the n about the gap mustbe used to alte the gap. f the n is not actually used in g the gap, then

e is no feedback.Fo the s of this , we have taken a d view of what constitutes

feedback, than exclude t evidence.One of the most t s of the effectiveness of feedback was d out

by & i (1996). They d ove 3000 s of the effects of feedbackon e (2500 s and 500 technical . Afte excluding thosewithout adequate , those e the feedback s e confoundedwith othe effects, e fewe than 10 s e included in the study, e

e was only discussed than , and those e insufficientdetails e given to estimate effect sizes, they e left with 131 , yielding 607effect sizes, and involving 12,652 .

They found an e effect size of 0.4 (equivalent to g the achievement ofthe e student to the 65th , but the d deviation of the effect sizeswas almost 1, and d two in y five effects e negative. The fact that so many

h s found that feedback can have negative effects on e suggeststhat these e not y s of poo design, o y in the , but

, substantive effects.n to explain the y in the d effect sizes, they examined possible

' of the effectiveness of feedback t is s whichimpact, eithe negatively o positively, on the effectiveness of feedback.

They began by noting that d with a 'gap' between actual and e levelsof some e (what & , 1996, m a d

, e e fou d classes of action.The t is to attempt to h the d o e level, which is the typical

e when the goal is , e the individual has a high commitment toachieving the goal and e the individual's belief in eventual success is high. Thesecond type of e is to abandon the d completely, which is y

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 49

common e the individual's belief in eventual success is low (leading to d, 1986). A , and less , e is to change the

, than abandoning it . s may lowe the ,especially likely e they cannot o do not want to abandon it, and , may,if successful, choose to e the . The h e to a dgap is simply to deny it exists.

& i (1996) found l t fo each of these s of, and developed a l model that accounted fo a significant n

of the y in effect sizes found in the . They identified e levels oflinked s involved in the n of task : meta-task processes:,involving the self; task-motivation processes, involving the focal task; and task learningprocesses involving the details of the focal task.

n g a typology of teache feedback, based on m , Tunstall &Gipps (1996b) d thei s types in a , g m those that

t attention to the task and to g methods, to those which t attentionto the self—in the e s by s only on s and punishments. These

s did not study effects on , but such studies by s (e.g. o & vanOudenhoven, 1995) show that feedback s that cue individuals to tattention to the self than the task appea to be likely to have negative effects on

. Thus , like othe cues which w attention to self-esteem and awaym the task, y has a negative effect (and goes some way to explaining why

l studies, such as Good & s (1975), found that the most effective sactually e less than .

This may explain the s obtained by t et al. (1990). A p of 80 Canadianstudents in thei d yea of y schooling e y assigned to one of

e s fo a e on the g of the majo scales in music e e nos between the s in s of musical aptitude, s academic success

o ability to . g the e of thei , the l p(GE1) e given feedback on thei t in the m of n , a list ofweaknesses and a n fo , while the second l

p (GE2) e given l feedback, told about thei s and given they to t them. On the post-test, the second l p had

gained e than eithe the l p o the l p (which enot significantly . One n of this t is that the l y offeedback is e effective than n y of feedback. , it seems eplausible that the y message which d the n feedback cuedthe pupils into a focus on meta-task , than on the tasks themselves.

evidence of the negative effect of cueing pupils to focus on the selfthan the task comes m a study d out by (1987) in which she examinedthe effects of fou kinds of feedback (comments, , , no feedback) on the

e of 200 i e 5 and 6 students in t thinking tasks. Althoughthe fou s e matched on t , the students given comments d

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 50 Black & D. Wiliam

Locus ofmotivation

External

Internal

Value system

External

Externalregulation

Introjectedregulation

Internal

Identifiedregulation

Integratedregulation

. 1. Classification of behaviou , based on i & n (1994).

one d deviation highe than the othe s on the post-test e e nosignificant s between the othe e . , sgiven to the students at the end of the sessions showed that the students given sand e d fa highe than the 'comments' o the 'no feedback' s on

s of ego-involvement while those given comments d highe than the othee s on s of task-involvement. , those given e had the

highest s of success, even though they had been significantly less successfulthan the 'comments' . This is consistent with the findings of n & e(1994), who found that while l e and e feedback can estudents' t in and attitude s a task, such feedback has little, if any, effecton .

These ideas e simila to the d by i & n (1994), whoidentify fou kinds of n of : , , identified and

. l n s s that e d by contingen-cies y l to the individual', (p. 6), while d n s to

s that e motivated by l s and s such as self-esteem-t contingencies' (p. 6). d n s when a behaviou o

n is adopted by the self as y t o valuable' (p. 6), althoughthe motivation is , while d n s the n ofidentified values and s into one's t sense of self (p. 6). These foukinds of n can e be d as the t of g the locus of thevalue system with that of the motivation, as shown in Fig. 1.

Within this , it can be seen that both l and l motivation canbe effective, but only when associated with , as opposed to , valuedaims. s fo g c motivation e discussed by Leppe & l(1989).

d to these findings is the e body of k on the way that students es fo success and , and in the k of k and he associates

(see , 1986 fo a . The l s appea to be:n (whethe the s e l o ;

e (whethe the s e stable o unstable);specificity (whethe the s e specific and isolated o whethe they e

global, e and .

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 51

The clea message the h on n y (see fo example Vispoel &Austin, 1995) is that s must aim to inculcate in thei students the idea thatsuccess is due to , unstable, specific s such as , than on stable

l s such as ability ) o whethe one is positively d by theteache .

Task

n t to those s that cue attention to meta-task , feedbacks that t attention s the task itself e y much e

successful. s et al. (1991 a) used meta-analysis to condense the findingsof 40 studies into the effects of feedback in what they called 'test-like' events (e.g.evaluation questions in d g , w tests at the end of ablock of teaching, etc.). This study has y been discussed in the section on thequality of feedback. As pointed out , it was found that g feedback in the

m of s to the w questions was effective only when students could not'look ahead' to the s e they had attempted the questions themselves what

s et al. (1991a), called g fo h availability')., feedback was e effective when the feedback gave details of the t

, than simply indicating whethe the student's answe was t ot (see also , 1994). g fo these two s eliminated

almost all of the negative effect sizes that s et al. (1991a) found, yieldinga mean effect size s 30 studies of 0.58. They also found that the use of s

d effect sizes, possibly by giving s e in, o by acting as eadvance s , the l to be . They concluded that the key ein effective use of feedback is that it must e 'mindfulness' in the student's

e to the feedback. Simila s by (1991, 1992) m thesefindings, but also show that it is t fo the l between successive tests to

, with the test g y afte the t , but that theeffectiveness of successive tests is d if the students do not feel successful on the

test. Anothe t finding in s k is that tests e gas well as sampling it, thus g the often quoted analogy that 'weighing thepig does not fatten it'.

Also discussed in the section on the Quality of feedback was Elawa & s(1985) study of 18 y school , e it was found that the s dueto being given specific comments on s and suggestions fo , dwith being given just , e as t as the s in achievement due toattainment—a significant finding given the well-attested e of s attainment in

g e success.

Task Learning

What is g g the e is how little attention has been paidto task s in looking at the effectiveness of feedback. The quality of thefeedback , and in , how it s to the task in hand, is .

Feedback s to be less successful in 'heavily-cued' situations such as e found

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 52 Black & D. Wiliam

in d n and d g sequences, and ye successful in situations g ' thinking such as d

tests and n s s et al., 1991b) o concept mappingd & Naidu, 1992). Why this might be so is not , but one clue comes

a study d out by Simmons & Cope (1993). n this study, s of , aged9-11, with little o no e of Logo , showed highe levels of

e (as d by the SOLO taxonomy) when g on angle and ns on pape than when g in a Logo , which the sd to the y of the immediate feedback given in the Logo t

to e l o l and ' .y & s (1993) study of two d e classes found that students given a

'scaffolded' n as much o as little help as they dthose students given a complete solution as soon as they got stuck, and e e ableto apply thei knowledge to , o only slightly , tasks. Simila s e

d by s & n (1991) fo students who had used booklets of adjunctquestions in to monito thei s in tackling e . gstudents' skills in asking fo and giving help also has t positive effects onachievement d & , 1990; , 1995).

, the kind of help is t too. Some s have found that dexplanation of techniques that have y led to e is less effective than using

e s (Fuchs et al., 1991), although y (1992) suggests that the se inconclusive on this point. e is also evidence that the quality of dialogue in a

feedback n is t et al., 1995) and can, in fact, be esignificant than ability and y s combined , 1988).

, while focusing on s goals leads to achievement gainsthan a focus on t goals, feedback d to s seems to be e effectivethan feedback on absolute levels of e (Schunk & , 1991; Schunk &

, 1993a).n all this, it is easy to gain the n that e assessment is a static s

of g the amount of knowledge y possessed by the individual, andfeeding this back to the individual in some way. , as the meta-analysis of Fuchs& Fuchs (1986) showed, the effectiveness depends y on the systematic analysisand use of feedback by . , the account by Lidz (1995) of the yand e of dynamic assessment (and y the k of Vygotsky and

) makes plain that e assessment is as much d withn (i.e. what someone can ) as with what they have y , and it

is only in n with the (and the ) that useful assessments canbe made.

Prospects for the Theory and Practice of Formative Assessment

t might be seen , and indeed might be anticipated as conventional, fo aw of this type to attempt a meta-analysis of the quantitative studies that have been

. The fact that this y seems possible s a n on this field of

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 53

. l studies which e based on meta-analyses have d usefull fo this . , these have been focussed on w aspects

of e , fo example the y of questioning. The value of theis is also in question because key aspects of the s studies that they

synthesise, fo example the quality of the questions being d at the t is d because most of the s e no evidence about these

aspects.l quantitative studies which look at e assessment as a whole do

exist, and some have been discussed above, although the numbe with adequate ande quantitative would be of the of 20 at most. , whilst

these e s within thei own s and , and whilst they showsome e and t in n to the g gains associated with

m assessment initiatives, the g s between the studies esuch that any amalgamations of thei s would have little meaning.

At one level, these s e obvious on casual inspection, because each studyis associated with a pedagogy, with its attendant assumptions about :one that in many cases has been d as the main element of the innovationunde study. e e howeve deepe : even e the h studiesappea to be simila in the s involved, they diffe in the e of the datawhich may have been collected—o . The fact that t g

s e often given no attention is one sign of the inadequate conceptualisationof the issues involved, indicating a need fo y building. m the evidence

d in this , it is clea that a t deal of y building still needs to takeplace in the a of e assessment, and we shall make suggestions below abouta basis fo this development.

An g , which we have y noted in an pape (Wiliam& , 1996), is that the m e assessment' is not common in theassessment . Such meaning as we have attached to the m e is also

d fo s by such s as m evaluation', dassessment', 'feedback', e evaluation' and so on.

Taking the t in the section on feedback, we , fo the sakeof simplicity, that the m feedback be used in its least e sense, to to any

n that is d to the of any action about that .This need not y be an l e (as, fo example, would be

d by & , 1996), no need e y be some ed against which the e is , let alone some method of

g the two. The actual e can be evaluated eithe in its own ,o by g it with a e . The n can eithe be in sof equality (i.e. these e the same o , as a distance (how fa t of—oindeed beyond—the d was it?) o as diagnosis (what do need to do to get

. Adopting the definition (although not the ) d by Sadle (1989),we would e that the feedback in any assessment s a formative function onlyin the latte case. n othe , assessment is e only when n ofactual and e levels yields n which is then used to alte the gap. AsSadle , f the n is simply , passed to a d y who lacks

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 54 Black & D. Wiliam

eithe the knowledge o the powe to change the outcome, o is too deeply coded (foexample, as a y e given by the ) to lead to e action, the

l loop cannot be closed' , 1989, p. 121). n such a case, while theassessment might be e in , it would not be e in function andin ou view this suggests a basis fo distinguishing e and summative functionsof assessment.

Gipps (1994, chapte 9) s attention to a m shift m a testing eto an assessment , associated with a shift m s to the assessmentof . , Shinn & Good (1993) e that e needs to be a mshift' in assessment, m what they call the t assessment m (and whatwe have e called summative functions of assessment) to what they call the

g ' y equivalent to what we e e calling thee functions of assessment). They e the distinction by the s

in the way that questions e posed in the two s along s dimensions (seeTable m Shinn & , 1992). Summative functions of assessment e

d with consistency of decisions s ) e s of students, sothat the g e is that meanings e d by t s ofassessment . A m fo the s of summative assess-ments is that exactly who will be making use of the assessment s is likely to be

. n , e functions of assessment e econsequences eithe fo ) small s of students (such as a teaching )o fo individuals.

The lack of y about the e distinction is e o less evidentin much of the . Examples can be found in the of s and books,notably in the USA, about e assessment, authentic assessment, oassessment and so on, e innovations e , sometimes with evidencewhich is d as an evaluation, with the focus only on the y of the 'assessments and the feasibility of the m k involved. What is often missingis a clea indication as to whethe the innovation is meant to e the m

e of t of , o the m e of g a evalid m of summative assessment, o both.

The Theoretical Basis

All that can be set out e e a few 'notes s a y of e assessment',which e d y because they may be a helpful aid to n on the k

d and y because they may be helpful in looking ahead to the implicationsof this .

Two key , to which e has y been made, e those of Sadle(1989) and Tittle (1994). Sadle built upon s notion of the gap betweenthe state d by feedback and the d state, emphasising that action will beinhibited if this gap is seen as y wide. e d that ultimately,the action to close that gap must be taken by the student—a student who automaticallyfollows the diagnostic n of a teache without g of its eo n will not . Thus self-assessment by the student is not an g

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 55

option o ; it has to be seen as essential. Given this, the n by a studentof his o he k can only be e if that student comes to e the svision of the subject . Some (e.g. , 1995) e that this can be doneby g objectives, but s (e.g. Claxton, 1995; Wiliam, 1994) e that thesedefinitions must n implicit if they e not to t .

A development of this y seems to call fo links to compatible g sand to s of the meta-cognition and locus of l of the .

Tittle's (1994) k emphasises e dimensions. The the epistemologyand s involved, can e both to positions held in n to g in ,and to the epistemology t to the subject matte . The eof the epistemology, and so of the meta-cognition involved, in (say) aesthetic

n of y will be y t m that fo (say) physics, and hence manys of e assessment will diffe between these two fields of . The

second dimension is the e evident one of the assessment ; it can bed e that in l of the studies d , little is said about the detail

of these, o about the distinctive effects of the subject matte involved.Tittle's d dimension s in the and , and she y

s the e of these. n n to students, this emphasis is d anddeveloped by s , but the s beliefs, about the subject ,about , and about the students and the class, must also be tcomponents in any model, if only because it is on the basis of these that s of

s 'gap' must be . Tittle also makes the t point that whilen conceptions of validity y (e.g. , 1989) s the value-laden

e of assessment , the actual e of those values is excluded, gthe n that one y y ) set of values is as good asany . Thus t conceptions of validity e no guide as to what 'ought' tobe going on, y a l k fo discussing what is going on.

This emphasis on the ethical and l aspects of assessment is a e of thee outlined by Aikenhead (1997). e s upon the k of s

(1971, p. 308) and n (1988) to e that n of assessment can fallwithin e s that e commonly d in the social sciences. One, the

, y links to the c emphasis in d testing.The second, the e , has to be adopted in e assessment,and this link s out the e of g a s e in nto that s expectations and assumptions about the m , togethewith his o he n of the task demand and of the a fo success. n the

, the c , one would seek a e of the wide sbeing , notably the t of the , and the choice between eitheselecting an elite o achieving excellence fo all. This m also calls into play theneed fo a e of the g goals (and of the assessment a h whichthey e ) which should ask whose s these goals e designed to

.Simila s motivate the l k d by s (1997) as

a t of a detailed study of the changes (ove a two-yea ) of the e ofa single middle-school mathematics teache in the way she d to students'

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 56 Black & D. Wiliam

s to he questions. , the s s tended to focus on theextent to which the student s d with the s expectations (what

s s 'evaluative' listening). Afte sustained n and discussion with the ove a d of l months, the s n placed g

emphasis on ' as opposed to the ' whichd the lessons ' listening). s the end of the

two-yea , e was a shift in the s , with a d moveaway clea lesson s and d g outcomes, and sthe n of potentially h mathematical situations, in which the teache is a

. t notably, in this d phase, the s own views of the subjectmatte being 'taught' developed and d along with that of the students

' listening). t is clea e that a commitment to the use of eassessment y entails a move away y notions of intelligence (Wolfet al., 1991).

and the Social SettingThese last two analyses g out a e which in ou view has been absenta t deal of the h we have . This is that all the assessment s

, at , social , taking place in social settings, conducted by, on andfo social . Guy u (1984) has used the m 'didactical 'to e the k of y implicit) expectations and s that eevolved between students and . A e of such s is thatthey e to delimit 'legitimate' activity by the . Fo example, in a m

e the s questioning has always been d to ' skills,such as the n of t , students may well see questionsabout ' o 'application' as illegitimate o even meaningless(Schoenfeld, 1985).

As Tittle's (1994) h emphasises, the 'opening moves' of s and studentsin the negotiation of such m s will be d by thei epistemological,psychological and pedagogical beliefs. Fo example, when a teache questions a student,the s beliefs will influence both the questions asked and the way that s e

. An t e e is the distinction between 'fit' and 'match' (von, 1987, p. 13). Fo example, a teache may set student s in solving

systems of simple equations. f students answe all the questions , the teache maywell conclude that the students have ' the topic, i.e. they assume that thestudents' g matches . , this is not the case. Foexample, when asked to solve the following two equations

3a = 24a + b=16

many students believe that it is impossible, saying things like ' keep getting b is 8, butit can't be because a is 8'. This is because in the examples d in mosttextbooks, each lette stands fo a t . The students' g is

e not a match but only a 'fit' with the . The p between fitand match depends y on the of the questions used by the , and

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 57

this, in n will depend on the s subject knowledge, thei s of ,and thei e of .

A study of seven d y school s examined the implicita that s used to e whethe students had ' something

s et ah, 1995). Afte studying and discussing video s and sof lessons, seven s of ' d which e d by all seven

, although they e d not as a static check-list, but as a sof potential clues to the level of the student's :

(1) changes in : students who had d e ' whilethose who had not d ;

(2) extension of a concept: students who have d something often take theidea on thei own initiative;

(3) making modifications to a : students who , spontaneously tmaking thei own modifications, while those who don't d imitate ofollow ;

(4) using s in a t context: students who have d aidea often t seeing the same s ;

(5) using : only students who e e of the 'big ' can t ae so that thinking up o using a t is taken as evidence of

;(6) ability to explain: students who have d something e usually able to

explain it;(7) ability to focus attention: e on a task is taken as a sign of .

t may be that some s e content with 'fits' than 'matches' because theye e of the possibilities fo students' conceptions that e t m thei

own. , it seems likely that most s e e of the benefits ofquestioning styles, but find that such s e difficult to implement in l

' , 1990). n this , compute e that enables s toe e and diagnostic feedback may have a e to play a et al., 1993;

Wiliam, 1997), although e is little evidence so fa about the actual benefits of such.

n , the student's s to questioning will depend on a host of .Whethe the student believes ability to be l o fixed will have a ginfluence on how the student sees a question—as an y to n o as a tto self-esteem , 1986). Even e the student has a ' as opposed to

' , the student's belief about what counts as 'academic ', 1988) will have a d impact on the 'mindfulness' with which that

student . The study of two middle-school s by h et al. (1992)cited in the section on t e found that a majo t to the validityof t n was the extent to which students could t meaningsfo the tasks they e set, and the extent to which s could t meaningsfo the students' . They also found that s used assessment s asif they gave n on what students knew, , in fact, they e bette

s of motivation and task completion.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 58 Black & D. Wiliam

e specifically, the actual context of the assessment can also influence whatstudents believe is . An example is a study of a e 5 y class let ah, 1995) e e was assessed in two ways—via a multiple choice testand with an assignment in which students had to design a d y

. n the multiple choice test, the students focused on the s , whilein the l task, students engaged in much e n and qualitativediscussion of thei . s most significantly, discussion amongst students ofthe t s focused much e y on the subject matte (i.e. )than did the (intense) n of s on the multiple choice test.

The actions of s and students e also ' , 1991) by thes of schools and society and typically knowledge is closely tied to the situation

in which it is t , 1997). Spaces in schools e designated fo specifiedactivities, and given the e attached to ' in most ,

' actions e as often d with establishing , and studentsatisfaction as they e with developing the student's capabilities e & ,1995; & , 1996). A w by k (1996) shows that students e

y d and thei k d if they use s of em thei l s outside school and File (1993) found that n

g g and spelling in English y school s ed by the teache to develop these skills in d contexts, so that thei own

l s e 'blocked out'. n this way, , y 'objective'assessments made by s may be little e than the t of successivesedimentation of s ' assessments—in e cases the self-fulfilling

y of ' labelling of students , 1995).n g to e these effects of e and agency, s notion of

habitus , 1985) may be y . l s tosociological analysis have used e s such as , , and social classto 'explain' s in, fo example, outcomes, thus tending to t all those withina y as being homogenous. u uses the notion of habitus to e the

, s and positions adopted by social , y into account fo the s between individuals in the same . Such a notionseems y e fo g , in view of the fact that the

s of students in the same m can be so t t & ,1989).

—-prospects and needs

The above discussion has clea implications fo the design of h investigations.t s attention to the e of t s which will combine to e

the effects of any m . n the light of such a specification, it is clea thatmost of the studies in the e have not attended to some of the t aspectsof the situations being . A full list of t and t aspects wouldinclude the following:• the assumptions about g g the m and pedagogy;• the e g the composition and n of the g ;

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 59

• the e e of the s types of assessment evidence d by thes ;

• the e used by both s and s in g to thisevidence;

• the g k used in acting on the s so ;• the divisions of y between s and s in these ;• the s and beliefs held by the s about themselves as s about

thei own g , and about the aims and methods fo thei studies;• the s and beliefs of s about , about the 'abilities' and

s of thei students, and about thei s as ;• the e of the social setting in the , as d by the g and

teaching s and by the s of the wide school system as they eand evaluate them;

• issues g to , class and , which appea to have d little attentionin h studies of e assessment;

• the extent to which the context of any study is l and the possible effects ofthis e on the y of the .

To make adequate t of all of these, let alone l them in any classicalquantitative design, would seem y difficult. This is not to imply that e

s of outcomes, both of g and of attitudes to the subjects , e notto be sought—although one of the s evident in many of the studies seems tobe that although they e g g aims that the established methods e oplay down, they have to justify themselves in n to tests which e adapted to theestablished methods only. e is y a need fo a combination of such swith qualitative studies of s and s within the . ,as we believe, e is a need to evolve new s as quickly as possible, suchstudies might well focus on the s of change and attendant .

attention ought to be paid to two specific . The is theevidence in many studies that new emphasis on e assessment is ofbenefit to the disadvantaged and low-attaining e which is not

d in the s of othe studies. The t s e ye because e e some t s of the s that have yet to be

d and . f it is e that the s of school achievement might bed by the enhancement of the achievement of those o seen as slow, then e e y g social and educational s fo giving high

y to sensitive h and development k to see how to d andtackle the issues involved.

The second , o clutch of , s to the possible confusions andtensions, both fo s and , between the e and summative

s which thei k might have to . t is inevitable that all will be involved,one way o the , in g to both , and if an optimum balance is notsought, e k will always be e because of the t of ddominance by the summative.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 60 Black & D. Wiliam

E . t questions g m Shift (Shinn & , 1992)

n t assessment m g m

e

Test Validity

Unit ofAnalysis

Time Line

Level ofe

Locus of them

Focus

Testy

Context

nofdependent

e

o assessment s d outindividuals facilitating classification/placement into ?

s the assessment device ewhat it says it ?

d Validity: sthe test e with othe tests

g to e the samething?

t Validity: s the testdisplay a stable facto ?

: c statementsabout individuals: o studentswith simila assessment smost likely display simila

?Summative: s the assessmentindicate whethe o not the

n did ?s the assessment e an

indirect e of an e?

s the assessment identifyt student characteristics that

e to m etiology?

m : sassessment y identifyproblems?

e test s stable ove time?

e s b a s e d o nbehavio samples, obtained in

t contexts/settingsconsistent?

s the assessment e an with students g

a nationally e e ofm and ?

s the assessment en g the level of

pupil ?

s assessment t in sociallymeaningful student outcomes fo theindividual?

e the inferences and actions basedon test s adequate and appropriate

, 1989)?t Validity: o decisions

g t s ands based on knowledge

obtained the assessmente t in better student

outcomes than decisions based one s s et al.,

1983)?: s assessment show

that this t is g fo thisstudent?

: s the assessmentindicate that this t is gfo this student?

s the assessment directly et t s o skills?

s assessment identify tcurriculum, instruction and contextual

s [that] e to msolution?

m Solution: s theassessment y identifysolutions?

What s account fo they in student ?

s the assessment e an with students ge m and?

s the assessment en g the level of

pupil e and the slope ofpupil ?

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 61

Are there for

Table could be d alongside the section on s and tactics fo s ashelping to e the essential elements of any y to e g h

h implementation of e assessment. These elements would be thesetting of clea goals, the choice, g and n of e g tasks,the deployment of these with e pedagogy to evoke feedback, noting the

s in the section on Students and e assessment, and the en and use of that feedback to guide the g y of students.

Within and g h any such plan should be a commitment to involvingstudents in the s of self- and t as emphasised in the abovesection, d by a t h to .

e e y many t ways in which such guidelines could be dinto m , and whilst the s s and schemes d

t this , and the s d in the section on Systems,give useful examples, e is y no single l . n , a l gof the sections on n and , on Goal n and Self-

n on students' e to feedback, and of s of sections on The qualityof feedback, Feedback and Expectations and the social setting, should show that in

g the feedback that they give to students s have to keep in mind lt and delicate s which e neithe widely known no .

Fo public policy s schools, the case to be made e is that significantg gains lie within ou . The h d e shows conclusively that

e assessment does e . The gains in achievement appea to bequite , and as noted , amongst the t eve d foeducational . As an n of just how big these gains , an effectsize of 0.7, if it could be achieved on a nationwide scale, would be equivalent to gthe mathematics attainment e of an ' y like England, New Zealando the United States into the 'top five' afte the c m s of ,

, Japan and g g n et ah, 1996).f this point is accepted, then the second move is fo s in schools to be

d and d in g to establish new s in e assessment,e being extensive evidence to show that the t levels of e in this aspect

of teaching e low , 1993b; m et al., 1993), and that the level ofs devoted to its , at least in the U since 1988, has been almost

negligible , 1995).e is no doubt that, whilst building t , adequate , and

d guides to , fo e assessment is a e, e is enough evidence in place fo giving helpful guidance to l

action (fo an account of a majo state-wide assessment system which se and summative functions of assessment, see, fo example, e & ,

1996). , despite the existence of some l and even negative ,the e of conditions and contexts unde which studies have shown that gains canbe achieved must indicate that the s that e achievement of substantial

s in g e . Significant gains can be achieved by many

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 62 Black & D. Wiliam

t , and initiatives e e not likely to fail h neglect of delicate andsubtle .

This last point is y t because e does not , m this t, any one optimum model on which such a policy might be based. What doese is a set of guiding , with the l caveat that the changes in

m e that e needed e l than , and have to bed by each teache into his o he e in his o he own way t

et al., 1996). That is to say, m in this dimension will inevitably take a long time,and need continuing t m both s and .

Acknowledgements

The compilation of this w was made possible by the n of a t theNuffield Foundation, whose t we y acknowledge. We would also liketo thank the s of the h Educational h Association's Assessment

y Task p who commissioned the , d helpful comments on, and made suggestions fo additional s fo inclusion. e such

, this w is bound to contain , omissions and ,which , of , the e y of the .

s, C , , . & , V. (1990) Assessment and teache autonomy, Cambridge

Journal of 20, pp. 123-133.Aikenhead, G. (1997) A fo g on assessment and evaluation, in: Globalization

of Science —-papers for the Seoul Conference, pp. 195-199 (Seoul, ,n Educational t .

, . (1995) An examination of the p between teache efficacy andd t and student-achievement, and Special 16,

pp. 247-254., C. (1992) : goals, , and student motivation, Journal of

84, pp. 261-271., C. & , J. (1988) Achievement goals in the : students' g s andmotivation , Journal of 80, pp. 260-267.

, . (1995) Student self-evaluations—how useful—how valid, Journal of Studies, 32, pp. 271-276.

, . & , G.O. (1994) y ' assessment s as din the e of h Columbia, Canada, Assessment in 1, pp. 63-93., , , , GUNSTONE, . & , . (1991) The e of nin g science teaching and , Journal of in Science Teaching, 28,pp. 163-182.

, , C.-L.C., , J.A. & , . (1991a) The leffect of feedback in test-like events), of 61, pp. 213-238.

, , , J.A. & , C.-L.C. (1991b) Effects of mtesting, Journal of 85, pp. 89-99.

, A.E., , , , , GONZALEZ, E.J., , . & , T.A. (1996) Achievement in the School Years , , n College).

, . & , . (Eds) (1991) process and product , ,.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 63

, S.N., , E.C., , C.G. & , . (1992) A longitudinal study ofy s d competence in, and s about, national m

implementation, in 7, pp. 53-78,, J. & YANG, . (1996) A n of o assessment based on : findings and

implications m a e scale , Journal of and Development in 29,pp. 181-191.

, , , , , . & , A.N. (1991) Effects of a t andplanning system on ' cognitive development and educational ,American Journal, 28, pp. 683-714.

, . & , S. (1992) , concept mapping and feedback—a distanceeducation field , British Journal of Technology, 23, pp. 48-60., . & , . (1984) Criterion-referenced assessment in the classroom ,

Scottish Council fo h in Education)., . & , . (1996) Changing the subject: innovations in science, mathematics and technologyeducation (London, e with ., . (1993a) Assessment policy and public confidence: comments on the A y Task

s e 'Assessment and the t of education', The Curriculum Journal, 4,pp. 421-427., . (1993b) e and summative assessment by , Studies in Science

21, pp. 49-97., . & , G. (1990) , School Science 71(255), pp. 142-144., . & , . (1976) y , in: L.S. N (Ed.) of in

, , .. (1992) m g and motivation: g and expanding goal ,

Journal of 84, pp. 272-281., J. (1997) school mathematics: teaching styles, sex and setting , U

Open y ., L. & , A. (1996) The n between ' l goals and thei

assessment s in high school biology , Science 80, pp. 145-163., J.J. (1991) The mechanisms g the g s of pupils: n to a

y of e assessment, in: . WESTON (Ed.) Assessment of Achievement- and School Success, pp. 119-137 , Swets and .

, , , G. & , . (1990) e evaluation effects on g music,Journal of 84, pp. 119-125.

, . (1985) The genesis of the concepts of 'habitus' and 'field', Sociocriticism, 2(2),pp. 11-24.

. (1992) l evaluation: a case study of the national evaluation of s ofachievement ) , British Journal, 18, pp. 245-260.

, , , , , S., NUTTALL, . & , . (1990) s ofachievement: t of the National Evaluation of t Schemes, in: T. N (Ed.)Assessment Debates, pp. 87-103 (London, and Stoughton).

, , , , , C. & , A. (1996) Assessment in h yschools, Curriculum Journal, 7, pp. 227-246.

, . & , . (1994) e development of subject matte in mathematics, Studies in 27, pp. 217-248.

, G. (1984) The l e of the didactical t in the analysis and nof situations in teaching and g mathematics, in: . (Ed.) Theory of

5 topic area and miniconference, pp. 110-119 ,, t fu k de k de t .

, A.L., , J.C., , L.S. & , . (1992) e g: a new look at assessment and , in: . & . O'CONNO

(Eds) Changing Assessments: alternative views of aptitude, achievement and instruction,pp. 121-211 n USA and t , .

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 64 Black & D. Wiliam

, . & , . (1987) The feasibility of class d diagnostic assessment iny mathematics, 29, pp. 95-107.

, , , , VAN , . & , T. (1996) A l validationof l s n (with low-achieving e , Journal of

88, pp. 18-37., . & , . (1995) Feedback and d : a l synthesis,

of 65, pp. 245-281., J. (1995) ' judging s in senio science subjects: fifteen s of the

Queensland , Studies in Science 26, pp. 135-157., J. & , W. (1987) The impact of assessment changes on the science ,

in Science 17, pp. 236-243., . (1987) Task-involving and ego-involving s of evaluation: effects of t

feedback conditions on motivational , t and , Journal of 79, pp. 474-482.

, . (1988) Enhancing and g c motivation; the effects of task-involving andego-involving evaluation on t and , British Journal of58, pp. 1-14.

, . & , O. (1995) Effects of task and ego-achievement goals on help-seekings and attitudes, Journal of 87, pp. 261-271.

CALFEE, . & , S.W. (1996) m g : old, new, , blue,in: . CALFEE & . O (Eds) Writing in the Classroom, pp. 3-26 , NJ,

e .CALFEE, . & , . (Eds) (1996a) Writing in the Classroom , NJ,

e .CALFEE, . C. & , . (1996b) A national y of o : what we d and

what it means, in: . CALFEE & . O (Eds) Writing in the Classroom, pp. 63-81, NJ, e .

, J. & , . (1994) , , and c motivation: ameta-analysis, of 64, pp. 363-423.

, W.S. (1991) Questioning in a sociolinguistic , of 61, pp. 157-178.

, . (1994) Using d examples as an l t in the a ,Journal of 83, pp. 360-367.

, S., GALTON, , , L. . (1990) Observing Activities (London,l Chapman).

, G.J., , . & , . (1995) ' assessment : ,isolation and the kitchen sink, Assessment, 3, pp. 159-179.

, J. (1988) m dialogue and science achievement, in Science 18,pp. 83-94.

CLAXTON, G. (1995) What kind of g does self-assessment ? g a 'nose' foquality; comments on , Assessment in 2, pp. 339-343.

, A. (1992) s fo science education: issues in , e and authenticity,Science 76, pp. 451-463.

, . & , L. (1996) ' s and g science andtechnology, Journal of Science 18, pp. 105-116.

, . & , . (1993) Assessment in Higher politics, pedagogy, andportfolios , CT, .

, , , . & , . (1991) Effects of y focused feedback onenhancement of academic self-concept, Journal of 83, pp. 17-27.

, T.J. (1988) The impact of m evaluation s on students, of 58, pp. 438-481.

, . (1996) s and o assessment, in: . N & . WOLF (Eds) Student Assessment: challenges and possibilities, pp. 239-260 (Chicago, ,

y of Chicago .

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 65

, . & , J.A. (1989) t students in yea 8 science : a n withand extension of existing , in Science 19, pp. 67-75., C. (1990) m a l to a l method of g educational diagnosis with

m assessment, Alberta Journal of 36, pp. 35-44., C., J. & , . (1993) e assessment in a m setting: m

e to compute innovations, Alberta Journal of 39, pp. 111-125., . (1995) Curriculum Assessment. A review of policy 1987-1994 (London,

Falme ., . (1997) Listening fo : an evolving conception of mathematics teaching, Journalfor in 28, pp. 355-376., N . & , . (1996) e assessment: to what extent is its potential to enhance pupils'science being ? School Science 77(281), pp. 93-100.

, . & , L.A. (1993) Static and dynamic s of ability: an l, Journal of 85, pp. 76-82.

, E.L. & , . (1994) g d education, Scandinavian Journal of 38, pp. 3-14.

, . & , C. (1991) Effects of y g and e non s g , Journal of 83,pp. 35-42.

, F. N. (1991) Synthesis of h on s and tests, Leadership, 48(7),pp. 71-76.

, F.N. (1992) Using tests to e : a neglected m , Journalof and Development in 25, pp. 213-217., S.L. (1993) d , in: J.J. (Ed.) Curriculum-Based

pp. 1-24 (Lincoln, NE, s e of l ., W. (1988) k in mathematics classes: the context of students' thinking g ,

23, pp. 167-180., A. & , C. (1987) The stepping stones of g and evaluation,

Journal of Science 9, pp. 93-104., . & , . (1997) s and challenges to changing the focus of

assessment and n in science , Assessment, 4, pp. 37-73., C.S. (1986) l s affecting , American (Special

science and education), 41, pp. 1040-1048.. (1988) g in assessment, School Science 70(252), pp. 119-125.

, . & SUTTON, A. (1991) A l h to d , British Journalof Technology, 23, pp. 4-20.

, . & , L. (1985) A l t in ' n feedback on student: changing teache behaviou a little than a lot, Journal of

77, pp. 162-173., . (1994) Feedback in , 26, pp. 58-73.

, , , . & , . (Eds) (1994) Teachers Assessing lessons from scienceclassrooms , Association fo Science Education).

, . & FONTANA, . (1996) Changes in l beliefs in e y schoolpupils as a consequence of the employment of self-assessment , British Journal of

66, pp. 301-313., . & , (1989) OCEA The development of a t scheme in science,t The pilot phase, School Science 70(253), pp. 97-102.

, A. (1993) Contexts of assessment in a y , BritishJournal, 19, pp. 95-107., A. (1995) Teache Assessment: social s and social , Assessment in2, pp. 23-38.

FONTANA, . & , . (1994) s in mathematics e as aconsequence of self-assessment in e y school pupils, British Journal of

64, pp. 407-417.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 66 Black & D. Wiliam

Foos, , , J.J. & , S. (1994) Student study techniques and the n effect,Journal of 86, pp. 567-576.

, . & , . (1997) e assessment of students' h within and middle school science , pape d at the Annual of the

Chicago 1997., L.S. (1993) Enhancing l g and student achievement with

d , in: J.J. (Ed.) Curriculum-Basedpp. 65-103 (Lincoln, NE, s e of l ., L.S. & , . (1986) Effects of systematic e evaluation: a meta-analysis,

Children, 53, pp. 199-208., L.S., , , , C.L. & , . (1991) Effects of d

t and consultation on teache planning and student achievement in mathematics, American Journal, 28, pp. 617-641.

, L.S., , , , C.L., WALZ, L. & , G. (1993) e evaluationof academic w much h can we expect? School 22,pp. 27-48.

, . & , . (1989) Teaching fo : y e in high school, Journal of in Science Teaching, 26, pp. 1-14.

, E. & , . (1995) The d y of g : amulti-faceted e on individual s in , in: . & .

Y (Eds) Alternatives in Assessment of Achievements, Learning andpp. 283-317 , .

, G. (1996) g an assessment stance in y education in England, Assessmentin 3, pp. 55-74., C. (1994) Beyond Testing: towards a theory of educational assessment (London, Falme ., C., , . & , . (1997) s of teache assessment among y school

s in England, The Curriculum Journal, 7, pp. 167-183., T . L . & , . A. (1975) product relationships in fourth-grade mathematics classrooms(Columbia, , y of .

, A.C., , . & , . (1995) e dialogue s inc one-to-one , Applied Cognitive 9, pp. 495-522.

, . & , G. (1993) g to : action h m an equal se in a junio school, British Journal, 19, pp. 43-58.

, A. (1991) g assessment in y schools: ' h s e ,in: . WESTON (Ed.) Assessment of Achievement: motivation and school success, pp.103-118 , Swets and .

, W.S. & , . (1987) Autonomy in s : an land individual e investigation, Journal of and Social 52,pp. 890-898.

, . & GATES, S.L. (1986) Synthesis of h on the effects of y g iny and y , Leadership, 33(8), pp. 73-80.

Guskey, . & , . (1988) h on d y g : ameta-analysis, Journal of 81, pp. 197-216.

, J. (1971) and Human , , ., , , J. & . (1995) A case study of systemic aspects of assessmenttechnologies, Assessment, 3, pp. 315-361., , , , , S., YOUNG, V. & , . (1997) A study of teache assessmentat key stage 1, Cambridge Journal of 27, pp. 107-122.

, W., , C , , . & NUTTALL, . (1992) Assessment and the tof education, The Curriculum Journal, 3, pp. 215-230.

, W., , . & , . (1995) ' assessment and national testing iny schools in Scotland: s and , Assessment in 2, pp. 126-144.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 67

, W. & , . (1996) Assessment and testing in Scottish y schools, TheCurriculum Journal, 7, pp. 247-257.

, . (1993) g n in s : the use of visual sto assess y school s g in , Cambridge Journal of 23,pp. 137-154.

, J., , J. & . (1996) Effects of g skills s on student :a meta-analysis, of 66, pp. 99—136., S.C., NELSON, . . (1983) The t utility of assessment: a functional

h to evaluating assessment quality, American 62, pp. 963-974., J.L., , . & , . (1996) s fo m assessment:

design and implementation issues, in: . CALFEE & . O (Eds) Writing inthe Classroom, pp. 27-59 , NJ, e .

, , , N.A. & , L.L. (1994) g assessment into the hands of young: a study of d a and self-assessment, Assessment, 2,

pp. 309-324., . & , . (Eds) (1997) Scaffolding Student Learning: instructional approaches and

issues , , ., W.G. & , G. (1991) Enhancing g using questions adjunct to science ,

Journal of in Science Teaching, 28, pp. 523-535., W.G. & , . (1992) w can n adjunct questions focus students

attention and enhance g of a d science lesson? Journal of Science Teaching, 29, pp. 3-16.

, . & , . (1993) Staff and p assessment of l communication skills,Studies in Higher 18, pp. 379-385.

, C. (1990) ' attitudes s GAS d Assessments in Science ,School Science 72(258) pp. 133-137.

, , , G.L. & , L.E. (1994) , d testing as an l, Journal of 62, pp. 93-101.

, . (1990) Negotiation and dialogue in student assessment, in: T. N (Ed.) AssessmentDebates, pp. 104-115 (London, and Stoughton).

, . & , . (1990) e g and achievement, in: S. N(Ed.) Co-operative Learning: theory and research, pp. 23-27 (New , .

, , , S., , , , J. & . (1995) Assessment of teachingand g in d , Teaching and Teacher 11,pp. 359-371.

, A.S.S. & , G.S. (1992) The impact of m testing y on high-school-students' achievement, Contemporary 17, pp. 71-77.

, A. (1990) Enhancing pee n and g in the m h lquestioning, American Journal, 27, pp. 664—687.

, A. (1991) Effects of g in c questioning on s g, Journal of 83, pp. 307-317.

, A. (1992a) n of self-questioning, , and note-taking w as sfo g m , American Journal, 29, pp. 303—323.

, A. (1992b)/ Facilitating e g h guided d questioning, 27, pp. 111-126.

, A. (1994) Autonomy and question asking—the e of l l in guidedd questioning, Learning and Differences, 6, pp. 163-185.

, A. (1995) g minds y do want to know—using questioning to teach lthinking, Teaching of 22, pp. 13-17.

, A. & , . (1993) Effects of guided e questioning on sknowledge , Journal of 61, pp. 127-148.

, V. (1995) Student self-evaluation s in d teaching and gcontexts of a and England, Assessment in 2, pp. 145-163.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 68 Black & D. Wiliam

, A.N. & , A. (1996) The effects of feedback s on : a l, a meta-analysis, and a y feedback n , Bulletin,

119, pp. 254-284., A. & , G.E. (1991) t of g n of physics texts bystudents' question , Journal of Science 13, pp. 473—485.

, J.J. (Ed.) (1993) Curriculum-Based (Lincoln, NE, s e of l.

, C.-L.C. & , J.A. (1987) y testing and student : a meta-analysis, Journalof Technology Systems, 15, pp. 325-345. J.A. & , C.-L.C. (1989) s in education, Journal of

13, pp. 221-340., C.-L.C, , J.A. & , . (1990) Effectiveness of g

: a meta-analysis, of 60, pp. 265-299.LAN, W.Y., , L. & , G. (1994) The effects of a g s on college

students' g in an y statistics , Journal of 62,pp. 26-40.

, & , . (1989) c motivation in the , in: C. S & . S(Eds) on in the Classroom, Vol. 3, pp. 73-105 (San , CA, Academic

.Lidz, C.S. (1995) c assessment and the legacy of L.S. Vygotsky, School

16, pp. 143-153., J.A. & , . (1996) y g and the g y hypothesis,

Journal of 90, pp. 67-74., . & , . (1988) 0CEA—The development of a d assessment scheme inscience t School s 1986, School Science 70(252), pp. 103-112., . & , T. (1990) g , skills and n asessments—studentsystems, School Science 71(255), pp. 145-150.

, A .W., , , , C. & , S.U. (1992) An n of assessmentmethods in middle school science, Journal of Science 14, pp. 305-317.

, . & Fox, . (1991) o y findings on test expectancy e tom outcomes? of 61, pp. 94-106.

. & , . (1991) Effect of g on subsequent s in academicachievement tests, 33, pp. 151-154.

, . & , N.C. (1992) g d testing and teache effects ina l mathematics , British Journal of 62, pp. 356-363.

, G.N. & EVANS, J. (1986) A sense of n in n d assessment, Studiesin 12, pp. 257-265.

, Y. (1996) m assessment in k y schools, The CurriculumJournal, 7, pp. 259-269.

, , , S., , C. & , . (1993) Teache assessment at y Stage1, in 8, pp. 305-327.

. & , E.S. (1992) A n of , , andg with d t in g among students with g

disabilities, Journal of Special 26, pp. 162-180., . (1969) s influencing , of 39,

pp. 293-318., J. & , F. (1992) m management fo t : an application fo

e , Studies, 18, pp. 3-10., S. (1989) Validity, in: . N (Ed.) d edn, pp. 12-103

(London, Collie ., . & , E. (1997) Consensually n explanation in science teaching, Science

80, pp. 173-192.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 69

, . & , . (1989) Enhancing student achievement h computed , Journal of in Science Teaching, 26, pp. 567-573.

, . (1996) Statewide o Assessment: the t , in: . N & .WOLF (Eds) Student Assessment: challenges and possibilities, pp. 192-214(Chicago, , y of Chicago .

, T. (1991) Colonising y of a ., , , , STOLL, L. & , . (1988) School the junior years

, Open ., . (1992) The use of l feedback in s fo e

, Technology and Development, 40(3) pp. 5-20., G. (1987) The impact of evaluation s on students, 22,

pp. 155-175., . (1992) The assessment of e in social studies, in: . et al. (Eds)

Toward a Science of Testing and Assessment, pp. 53-69 (Albany, NY, Statey of New k .

, . & , . (1995) Students' help seeking g m solving: effectsof , goal, and achievement, American Journal, 32,pp. 352-376.

, . (1994) A k fo developing cognitively diagnostic assessments, of 64, pp. 575-603.

, J .C. & , . (1990) n evaluation and n in g mscience texts, Journal of in Science Teaching, 27, pp. 447-460.

, . & , . (1986) The impact of the d test movement on mteaching and , Studies in 12, pp. 275-279.

, . (1991) s a c h to e evaluation, in: . WESTON (Ed.)Assessment of Achievement: and School Success, pp. 79-101 ,Swets and .

, . (Ed.) (1997) Handbook of Classroom Assessment (San o CA and London, Academic.

, S.J. (1992) s in g student , of 38,pp. 117-131.

, A., , , , , , . & , . (1994) Changing Schools? The of the Act at Stage One (London, Cassell).

, . & , . (1994) Enabling pupils with g difficulties to t on thei ownthinking, British Journal, 20, pp. 579-593.

, , , E., , V.E., , V., , A. & , . (1992) gmindful use of knowledge—attempting to t y s facilitates

, 27, pp. 91-109., J. & , . (1996) l n in e assessment: assessing the

k o g the child? The Curriculum Journal, 7, pp. 205-226.. . (1996) l s in the use of s fo d ,

American Journal, 33, pp. 845-871., J. & , C. (1994) Teaching the language of : s a metacognitive

h to pupil , British Journal, 20, pp. 429-145., . (1994) The s of facilitating qualitative e assessment in pupils, British

Journal of 64, pp. 145-160., A. (1983) On the definition of feedback, Behavioral Science, 28, pp. 4-13.

, . (1992) The implementation of d assessment in the teaching ofscience, in Science and Technological 10, pp. 171-185.

, S., , . & X J . (1995) s of , Assessment,3, pp. 363-371.

, . (1996) The likelihood of success g m , ScandinavianJournal of 40, pp. 57-68.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 70 Black & D. Wiliam

, S. & , . (1995) Chemically speaking: a n of student-teache talk gy lessons using and building on students' , Journal of Science

17, pp. 797-809., , , C. & , S. (1996) Teaching students to e questions: a

w of the n studies, of 66, pp. 181-221., J.A. (1995) Effects of feedback on student behaviou in e g s in a

7 math class, School Journal, 96, pp. 125-143., , , , , S. & , C. (1993) Assessing Achievement in the s

, Open y ., . & , A. (1994) Science e h e conceptmapping: new s fo the , Journal of Science 16,pp. 437-455., . & , . (1996) Assessing, g and g students' educational :the case fo 'subject , Assessment in 3, pp. 309-351.

, . (1987) Testing and teaching: two sides of the same coin? Studies in 13, pp. 73-90.

, , A. . (1995) s on the implementation of Nationalm Science y fo the 5-14 age : findings and s m a national

evaluation study in England, Journal of Science 17, pp. 481-492., A.G. (1988) m evaluation within the : mapping the ,creation, diffusion, utilization, 10, pp. 25-47.

, . (1989) e assessment and the design of l systems,Science, 18, pp. 119-144., J. & , C. (1990) Curriculum Based Assessment (New , USA ., J. & , J.E. (1991) Assessment , , n .

, , , S. & , . (1992) t teaching, y , and yn with explicit : effects on the composition skills and self-efficacy of

students with g disabilities, Journal of 84, pp. 340-352., , , L., , W. & , T. (1990) Written Tasks (London, l

Chapman)., , , . & , . (1990) The impact of e and summative

assessment upon test e of science education , Teacher and Special 13, pp. 3-8.

, . (1985) problem-solving (New , NY, Academic .. (1996) Goal and self-evaluative influences g s cognitive skill ,

American Journal, 33, pp. 359-382., . & , . (1991) g goals and s feedback g g

n , Journal of Behaviour, 23, pp. 351-364., . & , C.W. (1993a) Goals and s feedback: effects on self-efficacy and

g achievement, Contemporary 18, pp. 337-354., . & , . (1993b) s feedback and goals, 15,

pp. 225-230.SCOTT, . (1991) s and themes: k and k assessment in the GCSE,

in 6, pp. 3-19., S.L., , C.E. & , . (1997) Assessment and g in high schoolmathematics , Journal for in 28, pp. 187-215.

, L.A (1995) Using assessment to e , Leadership, 52(5),pp. 38-43.

, L.A., , , , E.J., , S.F., , V. & WESTON, TJ. (1994)Effects of g m e assessments on student , in:of Annual of Conference, New : Available m C E 390918.

, L.A., , , , E.J., , S.F., , V. & WESTON, T.J. (1996)

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 71

Effects of g m e assessments on student , and 15, pp. 7—18.

, . & GOO , . (1993) : an assessment of its t status and s foits , in: (Ed.) Curriculum-Based pp. 139-178 (Lincoln, NE,

s e of l ., . & , . (1992) d t and gassessment—basic s and outcomes, Focus On Children, 24(5), pp. 1-20.

, E. (1995) Language Testing: matching assessment s with language knowledge,in: . & . Y (Eds) Alternatives in Assessment of Achievements, Learning

and pp. 143-160 , ., F. & VAN , . (1995) The effects of contingent feedback on d land , Journal of of 10, pp. 13-24.

, . & , . (1993) Angle and : effects of g types of feedback on thequality of , Studies in 24, pp. 163-176.

, . (1990) Why d assessment is unlikely to e , TheCurriculum Journal, 1, pp. 171-183.

, . (1990) n of d academic s and s with self-esteem insenio high school students, Scandinavian Journal of 34, pp. 259-269.

, T.F., , . & , S.L. (1997) t and dynamics of o assessment andl assessment in a college physics , Journal of in Science Teaching, 34,

pp. 255-271., . (1987) y g , of 57, pp. 175-214., . (1991) Synthesis of h on e , Leadership, 48(5),

pp: 71-82., , , N.A., , N.L., , L.J., , , , E., , ,

, , , . & , . (1992) Success fo all: a s h to nand y n in y schools. S h , VA, Educational

h ., , , N.A., , L. J., , , , S., , L. & , . (1996)

Success fo all: a y of , Journal of for Students at 1, pp. 41-76.

, J. (1991) p discussions in the , School Science 72(261), pp. 29-34., L.A.J. (1994) , self and tuto assessment: e , Studies in Higher

19, pp. 69-75., , , . & , . (1989) g thinking skills h

m assessment, Journal of 26, pp. 233-246., . (1989) The effect of testing on science s skill achievement, Journal of

in Science Teaching, 26, pp. 659-664., J. (1988) GAS The d assessments in science , School Science 70(251),pp. 152 -158., . (1989) The development of a k fo the assessment of s skills in a

d Assessments in Science , Journal of Science 11,pp. 251-259., . (1991) The e and assessment of scientific s in the , SchoolScience 72(260), pp. 65-76.

TAN, . (1992) An evaluation of the use of continuous assessment in the teaching of physiology,Higher 23, pp. 255-272.

. & , E. (1989) A meta-analysis of the effect of enhanced : cues,, t and feedback and s on moto skill , Journal of

and Development in 22(3), pp. 53-64., J.W. (1993) g independent g in the middle e e of

l t , School Journal, 93, pp. 575-591.

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 72 Black & D. Wiliam

, J.W., , L., , , , , , A. & , . (1993)s among students' study activities, self-concept of academic ability, and

achievement as a function of s of high-school biology , Applied Cognitive 7, pp. 499-532.

Tindal, G. (1993) A w of d s on nine assessment components, in: (Ed.) Curriculum-Based pp. 25-64 (Lincoln, NE, s e of

l ., . (1994) d an educational-psychology of assessment fo teaching and

, contexts, and validation , 29,pp. 149-162.

, . (1993) e assessment: some l s and l questions,Cambridge Journal of 23, pp. 333-343.

, . & , J. (1995) g teache assessment in infant :methodological s and g issues, Assessment in 2, pp. 305-320.

, . (1989) g achievement based assessment using e d ;in Science 19, pp. 286-290.

Tunstal l , . & , C. (1996a) w does you teache help you to make you k 's g of e assessment, The Curriculum Journal, 7, pp. 185-203.

TuNSTALL, . & , C. (1996b) Teache feedback to young n in e assessment: atypology, British Journal, 22, pp. 389-404.

, . & , . (1995) Success and e in junio high school: a l incidenth to g students' l beliefs, American

Journal, 32, pp. 377-412.VON , E. (1987) g as a e activity, in: (Ed.) of

in the Teaching and Learning of , NJ, e mAssociates)., . (1995) p n in assessment: multiple objectives, , andoutcomes, and Analysis, 17, pp. 239-261.

WESTON, C.J , L. & , T. (1995) A model fo g eevaluation in l design, and Development, 43(3),pp. 29-48.

, . & , A. (1992) The effects of pupil g ofon-task behaviou on y school , British Journal, 17,pp. 113-127.

, , VAN , J.W. & , G.F. (1995) y g in the , paped at the Annual of the San Francisco 1995, available m C

., . (1994) Assessing authentic tasks: s to , Studies in

2, pp. 48-68., . & , . (1996) s and consequences: a basis fo distinguishing e

and summative functions of assessment? British Journal, 22, pp. 537-548.Wiliam, . (1997, ) of the Stage 3 Diagnostic Software pilot, t d

fo School m and Assessment y (London, , s College LondonSchool of Education).

, G. (1987) m g y to assessment : a w of t n, Studies in 13, pp. 7-19.

WOLF, , , J., GLENN, J. & , . (Eds) (1991) To Use Their Well: investigatingnew forms of student assessment (Washington, , n Educational hAssociation).

YANCEY, . (1996) , y and : mapping the e and c of nin o assessment, in: . CALFEE & . O (Eds) Writing in the Classroom,pp. 83-102 , NJ, e .

ZESSOULES, . & , . (1991) Authentic assessment: beyond the d and into the

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 Assessment and Classroom Learning 73

, in: V. E (Ed.) Student Assessment, pp. 47-71 a ,USA, Association fo n and m .

, . & , A. (1994) t of y influences on g eattainment, American Journal, 31 , pp. 845-862.

, . & , . (1997) g a d : a social cognitive, Contemporary 22, pp. 73-101.

Appendix: list of journals searched

Fo the s listed below, the contents lists e scanned to identify t , as din the . s m s not in this list e found by othe means.

1. American Journal2. Applied Cognitive3. Assessment and in Higher4. Assessment in principles, policy and practice5. British Journal6. British Journal of Curriculum and Assessment7. British Journal of8. British Journal of Studies9. British Journal of Technology

10. Cambridge Journal of11. Child Development12. College Teaching13. Contemporary14. m 91)15. Technology and Development16. and17. Assessment18. and Analysis19. Leadership20. issues and practice21.22.23. m 92)24. Studies in25. Technology and Training26. School Journal27. in28.29. Journal of30. Journal of of31. Focus on Children32. Harvard33. Journal of (92-96)34. Journal of35. Journal of Studies36. Journal of Science37. of38. Journal for in39. Journal of )40. Journal of41. Journal of

Dow

nloa

ded

By:

[Sw

ets

Con

tent

Dis

trib

utio

n] A

t: 11

:29

8 M

ay 2

008 74 Black & D. Wiliam

42. Journal of43. Journal of and Behavioural Statistics44. Journal of (up to 94/5)45. Journal of Higher46. Journal of and Social (up to 93)47. Journal of48. Journal of Behaviour (up to 95)49. Journal of and Development in50. Journal of in Science Teaching51. Journal of School52. Journal of Special53. Learning and Differences m 91)54. Learning Disability in Focus (to 90)55. and m 92)56. Organizational Behaviour and Human Decision57. Oxford of58. Delta59. Bulletin60. in the Schools61. and Special62. in Science63. in64. of65. of in66. Scandinavian Journal of67. School68. Science69. Studies in70. Studies in Higher71. Studies in Science72. Teachers College73. Teaching and Teaching74. The Curriculum Journal75. Studies in76. Young Children